Spaces:
Runtime error
Runtime error
Commit
•
3dac99f
1
Parent(s):
70fea71
Upload 135 files
Browse filesThis view is limited to 50 files because it contains too many changes.
See raw diff
- GETTING_STARTED.md +67 -0
- INSTALL.md +24 -0
- README.md +133 -13
- app.py +202 -0
- configs/coco/Base-COCO-PanopticSegmentation.yaml +47 -0
- configs/coco/frozenseg/convnext_large_eval_a847.yaml +10 -0
- configs/coco/frozenseg/convnext_large_eval_ade20k.yaml +29 -0
- configs/coco/frozenseg/convnext_large_eval_bdd_panop.yaml +14 -0
- configs/coco/frozenseg/convnext_large_eval_bdd_sem.yaml +13 -0
- configs/coco/frozenseg/convnext_large_eval_cityscapes.yaml +8 -0
- configs/coco/frozenseg/convnext_large_eval_coco.yaml +3 -0
- configs/coco/frozenseg/convnext_large_eval_lvis.yaml +11 -0
- configs/coco/frozenseg/convnext_large_eval_mapillary_vistas.yaml +12 -0
- configs/coco/frozenseg/convnext_large_eval_pas21.yaml +10 -0
- configs/coco/frozenseg/convnext_large_eval_pc459.yaml +10 -0
- configs/coco/frozenseg/r50x64_eval_ade20k.yaml +13 -0
- configs/coco/maskformer2_R50_bs16_50ep.yaml +45 -0
- datasets/README.md +262 -0
- datasets/ade20k_instance_catid_mapping.txt +104 -0
- datasets/ade20k_instance_imgCatIds.json +0 -0
- datasets/prepare_ade20k_full_sem_seg.py +1004 -0
- datasets/prepare_ade20k_ins_seg.py +111 -0
- datasets/prepare_ade20k_pan_seg.py +499 -0
- datasets/prepare_ade20k_sem_seg.py +26 -0
- datasets/prepare_coco_semantic_annos_from_panoptic_annos.py +82 -0
- datasets/prepare_pascal_ctx_full_sem_seg.py +38 -0
- datasets/prepare_pascal_ctx_sem_seg.py +74 -0
- datasets/prepare_pascal_voc_sem_seg.py +55 -0
- demo/demo.py +189 -0
- demo/predictor.py +273 -0
- eval.sh +70 -0
- frozenseg/.DS_Store +0 -0
- frozenseg/__init__.py +26 -0
- frozenseg/config.py +132 -0
- frozenseg/data/.DS_Store +0 -0
- frozenseg/data/__init__.py +1 -0
- frozenseg/data/dataset_mappers/__init__.py +0 -0
- frozenseg/data/dataset_mappers/bdd_semseg_dataset_mapper.py +107 -0
- frozenseg/data/dataset_mappers/coco_instance_new_baseline_dataset_mapper.py +187 -0
- frozenseg/data/dataset_mappers/coco_panoptic_new_baseline_dataset_mapper.py +163 -0
- frozenseg/data/dataset_mappers/mask_former_instance_dataset_mapper.py +179 -0
- frozenseg/data/dataset_mappers/mask_former_panoptic_dataset_mapper.py +164 -0
- frozenseg/data/dataset_mappers/mask_former_semantic_dataset_mapper.py +183 -0
- frozenseg/data/datasets/__init__.py +18 -0
- frozenseg/data/datasets/ade20k_150_with_prompt_eng.txt +151 -0
- frozenseg/data/datasets/ade20k_847_with_prompt_eng.txt +848 -0
- frozenseg/data/datasets/cityscapes_with_prompt_eng.txt +19 -0
- frozenseg/data/datasets/coco_panoptic_with_prompt_eng.txt +201 -0
- frozenseg/data/datasets/coco_stuff_with_prompt_eng.txt +183 -0
- frozenseg/data/datasets/lvis_1203_with_prompt_eng.txt +1203 -0
GETTING_STARTED.md
ADDED
|
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
## Getting Started with FrozenSeg
|
| 2 |
+
|
| 3 |
+
This document provides a brief intro of the usage of FrozenSeg.
|
| 4 |
+
|
| 5 |
+
Please see [Getting Started with Detectron2](https://github.com/facebookresearch/detectron2/blob/master/GETTING_STARTED.md) for full usage.
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
### Inference Demo with Pre-trained Models
|
| 9 |
+
|
| 10 |
+
We provide `demo.py` that is able to demo builtin configs. Run it with:
|
| 11 |
+
```
|
| 12 |
+
python demo.py \
|
| 13 |
+
--input input1.jpg input2.jpg \
|
| 14 |
+
[--other-options]
|
| 15 |
+
--opts MODEL.WEIGHTS /path/to/checkpoint_file
|
| 16 |
+
```
|
| 17 |
+
The configs are made for training, therefore we need to specify `MODEL.WEIGHTS` to a model from model zoo for evaluation.
|
| 18 |
+
This command will run the inference and show visualizations in an OpenCV window.
|
| 19 |
+
|
| 20 |
+
For details of the command line arguments, see `demo.py -h` or look at its source code
|
| 21 |
+
to understand its behavior. Some common arguments are:
|
| 22 |
+
* To run on your __webcam__, replace --input files with --webcam
|
| 23 |
+
* To run on a __video__, replace --input files with --video-input video.mp4.
|
| 24 |
+
* To run on __cpu__, add `MODEL.DEVICE cpu` after `--opts`.
|
| 25 |
+
* To save outputs to a directory (for images) or a file (for webcam or video), use `--output`.
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
### Training & Evaluation in Command Line
|
| 29 |
+
|
| 30 |
+
We provide a script `train_net.py`, that is made to train all the configs provided in FrozenSeg.
|
| 31 |
+
|
| 32 |
+
To train a model with "train_net.py", first setup the corresponding datasets following [datasets/README.md](./datasets/README.md), download [SAM checkpoints](https://github.com/facebookresearch/segment-anything?tab=readme-ov-file#model-checkpoints) and save it under `pretrained_checkpoint/`.
|
| 33 |
+
then run:
|
| 34 |
+
```
|
| 35 |
+
python train_net.py --num-gpus 4\
|
| 36 |
+
--config-file configs/coco/frozenseg/convnext_large_eval_ade20k.yaml
|
| 37 |
+
```
|
| 38 |
+
|
| 39 |
+
The configs are made for 4-GPU training.
|
| 40 |
+
Since we use ADAMW optimizer, it is not clear how to scale learning rate with batch size.
|
| 41 |
+
To train on 1 GPU, you need to figure out learning rate and batch size by yourself:
|
| 42 |
+
```
|
| 43 |
+
python train_net.py \
|
| 44 |
+
--config-file configs/coco/frozenseg/convnext_large_eval_ade20k.yaml \
|
| 45 |
+
--num-gpus 1 SOLVER.IMS_PER_BATCH SET_TO_SOME_REASONABLE_VALUE SOLVER.BASE_LR SET_TO_SOME_REASONABLE_VALUE
|
| 46 |
+
```
|
| 47 |
+
|
| 48 |
+
To evaluate a model's performance without `OpenSeg Ensemble`:
|
| 49 |
+
```
|
| 50 |
+
python train_net.py \
|
| 51 |
+
--config-file configs/coco/frozenseg/convnext_large_eval_ade20k.yaml \
|
| 52 |
+
--eval-only MODEL.WEIGHTS /path/to/checkpoint_file \
|
| 53 |
+
TEST.USE_SAM_MASKS False
|
| 54 |
+
```
|
| 55 |
+
|
| 56 |
+
For using `OpenSeg Ensemble`:
|
| 57 |
+
1. generate SAM mask predictions (default saveing under `output/SAM_masks_pred`):
|
| 58 |
+
```
|
| 59 |
+
python save_sam_masks.py --data_name pc_val --sam_model vit_h
|
| 60 |
+
```
|
| 61 |
+
2. run with:
|
| 62 |
+
```
|
| 63 |
+
python train_net.py \
|
| 64 |
+
--config-file configs/coco/frozenseg/convnext_large_eval_ade20k.yaml \
|
| 65 |
+
--eval-only MODEL.WEIGHTS /path/to/checkpoint_file \
|
| 66 |
+
TEST.USE_SAM_MASKS True
|
| 67 |
+
```
|
INSTALL.md
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
## Installation
|
| 2 |
+
The codebases are built on top of [Detectron2](https://detectron2.readthedocs.io/tutorials/install.html).
|
| 3 |
+
|
| 4 |
+
|
| 5 |
+
### Dependencies and Installation
|
| 6 |
+
```bash
|
| 7 |
+
conda create --name frozenseg python=3.10 -y
|
| 8 |
+
conda activate frozenseg
|
| 9 |
+
conda install pytorch==2.3.1 torchvision==0.18.1 -c pytorch -c nvidia
|
| 10 |
+
|
| 11 |
+
# under your working directory
|
| 12 |
+
python -m pip install 'git+https://github.com/facebookresearch/detectron2.git'
|
| 13 |
+
pip install git+https://github.com/cocodataset/panopticapi.git
|
| 14 |
+
pip install git+https://github.com/mcordts/cityscapesScripts.git
|
| 15 |
+
|
| 16 |
+
git clone https://github.com/chenxi52/FrozenSeg.git
|
| 17 |
+
cd FrozenSeg
|
| 18 |
+
pip install -r requirements.txt
|
| 19 |
+
|
| 20 |
+
# compile CUDA kernel for MSDeformAttn
|
| 21 |
+
cd frozenseg/modeling/pixel_decoder/ops
|
| 22 |
+
sh make.sh
|
| 23 |
+
cd ../../../..
|
| 24 |
+
```
|
README.md
CHANGED
|
@@ -1,13 +1,133 @@
|
|
| 1 |
-
|
| 2 |
-
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# FrozenSeg: Harmonizing Frozen Foundation Models for Open-Vocabulary Segmentation
|
| 2 |
+
|
| 3 |
+
This repository is the official implementation of FrozenSeg introduced in the paper:
|
| 4 |
+
>[**FrozenSeg: Harmonizing Frozen Foundation Models for Open-Vocabulary Segmentation**](https://arxiv.org/abs/2409.03525)
|
| 5 |
+
|
| 6 |
+
|
| 7 |
+
## Abstract
|
| 8 |
+
|
| 9 |
+
>Open-vocabulary segmentation is challenging, with the need of segmenting and recognizing objects for an open set of categories in unconstrained environments. Building on the success of powerful vision-language (ViL) foundation models like CLIP, recent efforts sought to harness their zero-short capabilities to recognize unseen categories. Despite demonstrating strong performances, they still face a fundamental challenge of generating precise mask proposals for unseen categories and scenarios, resulting in inferior segmentation performance eventually. To address this, we introduce a novel approach, FrozenSeg, designed to integrate spatial knowledge from a localization foundation model (e.g., SAM) and semantic knowledge extracted from a ViL model (e.g., CLIP), in a synergistic framework. Taking the ViL model's visual encoder as the feature backbone, we inject the space-aware feature into learnable query and CLIP feature in the transformer decoder. In addition, we devise a mask proposal ensemble strategy for further improving the recall rate and mask quality. To fully exploit pre-trained knowledge while minimizing training overhead, we freeze both foundation models, focusing optimization efforts solely on a light transformer decoder for mask proposal generation – the performance bottleneck. Extensive experiments show that FrozenSeg advances state-of-the-art results across various segmentation benchmarks, trained exclusively on COCO panoptic data and tested in a zero-shot manner.
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
## Dependencies and Installation
|
| 14 |
+
See [installation instructions](INSTALL.md).
|
| 15 |
+
|
| 16 |
+
## Getting Started
|
| 17 |
+
See [Preparing Datasets](datasets/README.md).
|
| 18 |
+
|
| 19 |
+
See [Getting Started](GETTING_STARTED.md).
|
| 20 |
+
|
| 21 |
+
|
| 22 |
+
## Models
|
| 23 |
+
<table>
|
| 24 |
+
<thead>
|
| 25 |
+
<tr>
|
| 26 |
+
<th align="center"></th>
|
| 27 |
+
<th align="center" style="text-align:center" colspan="4"><a href="logs/testing/ade20k.log">ADE20K(A-150)</th>
|
| 28 |
+
<th align="center" style="text-align:center" colspan="3"><a href="logs/testing/cityscapes.log">Cityscapes</th>
|
| 29 |
+
<th align="center" style="text-align:center" colspan="2"><a href="logs/testing/mapillary_vistas.log">Mapillary Vistas</th>
|
| 30 |
+
<th align="center" style="text-align:center" colspan="2"><a href="logs/testing/bdd100k.log">BDD 100K</th>
|
| 31 |
+
<th align="center" style="text-align:center" colspan="2"><a href="logs/testing/a-847.log"> A-847 </th>
|
| 32 |
+
<th align="center" style="text-align:center" colspan="2"><a href="logs/testing/pc-459.log"> PC-459 </th>
|
| 33 |
+
<th align="center" style="text-align:center" colspan="2"><a href="logs/testing/pas-21.log">PAS-21 </th>
|
| 34 |
+
<th align="center" style="text-align:center" ><a href="logs/testing/lvis.log">Lvis </th>
|
| 35 |
+
<th align="center" style="text-align:center" colspan="3"><a href="logs/testing/coco.log">COCO <br> (training dataset)</th>
|
| 36 |
+
<th align="center" style="text-align:center">download </th>
|
| 37 |
+
</tr>
|
| 38 |
+
</thead>
|
| 39 |
+
<tbody>
|
| 40 |
+
<tr>
|
| 41 |
+
<td align="center"></td>
|
| 42 |
+
<td align="center">PQ</td>
|
| 43 |
+
<td align="center">mAP</td>
|
| 44 |
+
<td align="center">mIoU</td>
|
| 45 |
+
<td align="center">FWIoU</td>
|
| 46 |
+
<td align="center">PQ</td>
|
| 47 |
+
<td align="center">mAP</td>
|
| 48 |
+
<td align="center">mIoU</td>
|
| 49 |
+
<td align="center">PQ</td>
|
| 50 |
+
<td align="center">mIoU</td>
|
| 51 |
+
<td align="center">PQ</td>
|
| 52 |
+
<td align="center">mIoU</td>
|
| 53 |
+
<td align="center">mIoU</td>
|
| 54 |
+
<td align="center">FWIoU</td>
|
| 55 |
+
<td align="center">mIoU</td>
|
| 56 |
+
<td align="center">FWIoU</td>
|
| 57 |
+
<td align="center">mIoU</td>
|
| 58 |
+
<td align="center">FWIoU</td>
|
| 59 |
+
<td align="center">APr</td>
|
| 60 |
+
<td align="center">PQ</td>
|
| 61 |
+
<td align="center">mAP</td>
|
| 62 |
+
<td align="center">mIoU</td>
|
| 63 |
+
<td></td>
|
| 64 |
+
</tr>
|
| 65 |
+
<td align="center"><a href="configs/coco/frozenseg/r50x64_eval_ade20k.yaml"> FrozenSeg (ResNet50x64) </a></td>
|
| 66 |
+
<td align="center">23.1</td>
|
| 67 |
+
<td align="center">13.5</td>
|
| 68 |
+
<td align="center">30.7</td>
|
| 69 |
+
<td align="center">56.6</td>
|
| 70 |
+
<td align="center">45.2</td>
|
| 71 |
+
<td align="center">28.9</td>
|
| 72 |
+
<td align="center">56.0</td>
|
| 73 |
+
<td align="center">18.1</td>
|
| 74 |
+
<td align="center">27.7</td>
|
| 75 |
+
<td align="center">12.9</td>
|
| 76 |
+
<td align="center">46.2</td>
|
| 77 |
+
<td align="center">11.8</td>
|
| 78 |
+
<td align="center">52.8</td>
|
| 79 |
+
<td align="center">18.7</td>
|
| 80 |
+
<td align="center">60.1</td>
|
| 81 |
+
<td align="center">82.3</td>
|
| 82 |
+
<td align="center">92.1</td>
|
| 83 |
+
<td align="center">23.5</td>
|
| 84 |
+
<td align="center">55.7</td>
|
| 85 |
+
<td align="center">47.4</td>
|
| 86 |
+
<td align="center">65.4</td>
|
| 87 |
+
<td align="center"><a href=""> checkpoint </a></td>
|
| 88 |
+
</tr>
|
| 89 |
+
<tr>
|
| 90 |
+
<td align="center"><a href="configs/coco/frozenseg/convnext_large_eval_ade20k.yaml"> FrozenSeg (ConvNeXt-Large) </a></td>
|
| 91 |
+
<td align="center">25.9</td>
|
| 92 |
+
<td align="center">16.4</td>
|
| 93 |
+
<td align="center">34.4</td>
|
| 94 |
+
<td align="center">59.9</td>
|
| 95 |
+
<td align="center">45.8</td>
|
| 96 |
+
<td align="center">28.4</td>
|
| 97 |
+
<td align="center">56.8</td>
|
| 98 |
+
<td align="center">18.5</td>
|
| 99 |
+
<td align="center">27.3</td>
|
| 100 |
+
<td align="center">19.3</td>
|
| 101 |
+
<td align="center">52.3</td>
|
| 102 |
+
<td align="center">14.8</td>
|
| 103 |
+
<td align="center">51.4</td>
|
| 104 |
+
<td align="center">19.7</td>
|
| 105 |
+
<td align="center">60.2</td>
|
| 106 |
+
<td align="center">82.5</td>
|
| 107 |
+
<td align="center">92.1</td>
|
| 108 |
+
<td align="center">25.6</td>
|
| 109 |
+
<td align="center">56.2</td>
|
| 110 |
+
<td align="center">47.3</td>
|
| 111 |
+
<td align="center">65.5</td>
|
| 112 |
+
<td align="center"><a href="https://drive.google.com/file/d/1ThjVgY7nawm1AAP1LhrmGVlI3zr1EYMG/view?usp=drive_link"> checkpoint </a></td>
|
| 113 |
+
</tr>
|
| 114 |
+
</tbody>
|
| 115 |
+
</table>
|
| 116 |
+
|
| 117 |
+
|
| 118 |
+
|
| 119 |
+
## Citing
|
| 120 |
+
|
| 121 |
+
If you use FrozenSeg in your research, please use the following BibTeX entry.
|
| 122 |
+
|
| 123 |
+
```BibTeX
|
| 124 |
+
@misc{FrozenSeg,
|
| 125 |
+
title={FrozenSeg: Harmonizing Frozen Foundation Models for Open-Vocabulary Segmentation},
|
| 126 |
+
author={Xi Chen and Haosen Yang and Sheng Jin and Xiatian Zhu and Hongxun Yao},
|
| 127 |
+
publisher={arXiv:5835590},
|
| 128 |
+
year={2024}
|
| 129 |
+
}
|
| 130 |
+
```
|
| 131 |
+
|
| 132 |
+
## Acknowledgement
|
| 133 |
+
[Detectron2](https://github.com/facebookresearch/detectron2), [Mask2Former](https://github.com/facebookresearch/Mask2Former) and [OpenCLIP](https://github.com/mlfoundations/open_clip)
|
app.py
ADDED
|
@@ -0,0 +1,202 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import sys
|
| 3 |
+
|
| 4 |
+
# os.system("pip install gdown")
|
| 5 |
+
# os.system("pip install imutils")
|
| 6 |
+
# os.system("pip install gradio_client==0.2.7")
|
| 7 |
+
# os.system("python -m pip install 'git+https://github.com/facebookresearch/detectron2.git'")
|
| 8 |
+
# os.system("pip install git+https://github.com/cocodataset/panopticapi.git")
|
| 9 |
+
# os.system("python frozenseg/modeling/pixel_decoder/ops/setup.py build install")
|
| 10 |
+
|
| 11 |
+
import gradio as gr
|
| 12 |
+
from detectron2.utils.logger import setup_logger
|
| 13 |
+
from contextlib import ExitStack
|
| 14 |
+
import numpy as np
|
| 15 |
+
import cv2
|
| 16 |
+
import torch
|
| 17 |
+
import itertools
|
| 18 |
+
from detectron2.config import get_cfg
|
| 19 |
+
from detectron2.utils.visualizer import ColorMode, random_color
|
| 20 |
+
from detectron2.data import MetadataCatalog
|
| 21 |
+
|
| 22 |
+
from frozenseg import add_maskformer2_config, add_frozenseg_config
|
| 23 |
+
from demo.predictor import DefaultPredictor, OpenVocabVisualizer
|
| 24 |
+
from PIL import Image
|
| 25 |
+
import json
|
| 26 |
+
|
| 27 |
+
setup_logger()
|
| 28 |
+
logger = setup_logger(name="frozenseg")
|
| 29 |
+
cfg = get_cfg()
|
| 30 |
+
cfg.MODEL.DEVICE='cuda'
|
| 31 |
+
add_maskformer2_config(cfg)
|
| 32 |
+
add_frozenseg_config(cfg)
|
| 33 |
+
cfg.merge_from_file("configs/coco/frozenseg/convnext_large_eval_ade20k.yaml")
|
| 34 |
+
# os.system("gdown 1-91PIns86vyNaL3CzMmDD39zKGnPMtvj")
|
| 35 |
+
cfg.MODEL.WEIGHTS = './modified_model.pth'
|
| 36 |
+
cfg.MODEL.MASK_FORMER.TEST.SEMANTIC_ON = False
|
| 37 |
+
cfg.MODEL.MASK_FORMER.TEST.INSTANCE_ON = False
|
| 38 |
+
cfg.MODEL.MASK_FORMER.TEST.PANOPTIC_ON = True
|
| 39 |
+
predictor = DefaultPredictor(cfg)
|
| 40 |
+
|
| 41 |
+
|
| 42 |
+
title = "FrozenSeg"
|
| 43 |
+
article = "<p style='text-align: center'><a href='' target='_blank'>FrozenSeg</a> | <a href='' target='_blank'>Github Repo</a></p>"
|
| 44 |
+
|
| 45 |
+
examples = [
|
| 46 |
+
[
|
| 47 |
+
"demo/examples/ADE_val_00000001.jpg",
|
| 48 |
+
"",
|
| 49 |
+
["ADE (150 categories)"],
|
| 50 |
+
],
|
| 51 |
+
[
|
| 52 |
+
"demo/examples/frankfurt_000000_005898_leftImg8bit.png",
|
| 53 |
+
"",
|
| 54 |
+
["Cityscapes (19 categories)"],
|
| 55 |
+
]
|
| 56 |
+
]
|
| 57 |
+
|
| 58 |
+
|
| 59 |
+
coco_metadata = MetadataCatalog.get("openvocab_coco_2017_val_panoptic_with_sem_seg")
|
| 60 |
+
ade20k_metadata = MetadataCatalog.get("openvocab_ade20k_panoptic_val")
|
| 61 |
+
cityscapes_metadata = MetadataCatalog.get("openvocab_cityscapes_fine_panoptic_val")
|
| 62 |
+
lvis_classes = open("./frozenseg/data/datasets/lvis_1203_with_prompt_eng.txt", 'r').read().splitlines()
|
| 63 |
+
lvis_classes = [x[x.find(':')+1:] for x in lvis_classes]
|
| 64 |
+
lvis_colors = list(
|
| 65 |
+
itertools.islice(itertools.cycle(coco_metadata.stuff_colors), len(lvis_classes))
|
| 66 |
+
)
|
| 67 |
+
# rerrange to thing_classes, stuff_classes
|
| 68 |
+
coco_thing_classes = coco_metadata.thing_classes
|
| 69 |
+
coco_stuff_classes = [x for x in coco_metadata.stuff_classes if x not in coco_thing_classes]
|
| 70 |
+
coco_thing_colors = coco_metadata.thing_colors
|
| 71 |
+
coco_stuff_colors = [x for x in coco_metadata.stuff_colors if x not in coco_thing_colors]
|
| 72 |
+
ade20k_thing_classes = ade20k_metadata.thing_classes
|
| 73 |
+
ade20k_stuff_classes = [x for x in ade20k_metadata.stuff_classes if x not in ade20k_thing_classes]
|
| 74 |
+
ade20k_thing_colors = ade20k_metadata.thing_colors
|
| 75 |
+
ade20k_stuff_colors = [x for x in ade20k_metadata.stuff_colors if x not in ade20k_thing_colors]
|
| 76 |
+
cityscapes_stuff_classes = cityscapes_metadata.stuff_classes
|
| 77 |
+
cityscapes_stuff_color = cityscapes_metadata.stuff_colors
|
| 78 |
+
cityscapes_thing_classes = cityscapes_metadata.thing_classes
|
| 79 |
+
cityscapes_thing_color = cityscapes_metadata.thing_colors
|
| 80 |
+
|
| 81 |
+
def build_demo_classes_and_metadata(vocab, label_list):
|
| 82 |
+
extra_classes = []
|
| 83 |
+
|
| 84 |
+
if vocab:
|
| 85 |
+
for words in vocab.split(";"):
|
| 86 |
+
extra_classes.append(words)
|
| 87 |
+
extra_colors = [random_color(rgb=True, maximum=1) for _ in range(len(extra_classes))]
|
| 88 |
+
print("extra_classes:", extra_classes)
|
| 89 |
+
demo_thing_classes = extra_classes
|
| 90 |
+
demo_stuff_classes = []
|
| 91 |
+
demo_thing_colors = extra_colors
|
| 92 |
+
demo_stuff_colors = []
|
| 93 |
+
|
| 94 |
+
if any("COCO" in label for label in label_list):
|
| 95 |
+
demo_thing_classes += coco_thing_classes
|
| 96 |
+
demo_stuff_classes += coco_stuff_classes
|
| 97 |
+
demo_thing_colors += coco_thing_colors
|
| 98 |
+
demo_stuff_colors += coco_stuff_colors
|
| 99 |
+
if any("ADE" in label for label in label_list):
|
| 100 |
+
demo_thing_classes += ade20k_thing_classes
|
| 101 |
+
demo_stuff_classes += ade20k_stuff_classes
|
| 102 |
+
demo_thing_colors += ade20k_thing_colors
|
| 103 |
+
demo_stuff_colors += ade20k_stuff_colors
|
| 104 |
+
if any("LVIS" in label for label in label_list):
|
| 105 |
+
demo_thing_classes += lvis_classes
|
| 106 |
+
demo_thing_colors += lvis_colors
|
| 107 |
+
if any("Cityscapes" in label for label in label_list):
|
| 108 |
+
demo_thing_classes += cityscapes_thing_classes
|
| 109 |
+
demo_thing_colors += cityscapes_thing_color
|
| 110 |
+
demo_stuff_classes += cityscapes_stuff_classes
|
| 111 |
+
demo_stuff_colors += cityscapes_stuff_color
|
| 112 |
+
|
| 113 |
+
|
| 114 |
+
MetadataCatalog.pop("frozenseg_demo_metadata", None)
|
| 115 |
+
demo_metadata = MetadataCatalog.get("frozenseg_demo_metadata")
|
| 116 |
+
demo_metadata.thing_classes = demo_thing_classes
|
| 117 |
+
demo_metadata.stuff_classes = demo_thing_classes + demo_stuff_classes
|
| 118 |
+
demo_metadata.thing_colors = demo_thing_colors
|
| 119 |
+
demo_metadata.stuff_colors = demo_thing_colors + demo_stuff_colors
|
| 120 |
+
demo_metadata.stuff_dataset_id_to_contiguous_id = {
|
| 121 |
+
idx: idx for idx in range(len(demo_metadata.stuff_classes))
|
| 122 |
+
}
|
| 123 |
+
demo_metadata.thing_dataset_id_to_contiguous_id = {
|
| 124 |
+
idx: idx for idx in range(len(demo_metadata.thing_classes))
|
| 125 |
+
}
|
| 126 |
+
demo_classes = demo_thing_classes + demo_stuff_classes
|
| 127 |
+
return demo_classes, demo_metadata
|
| 128 |
+
|
| 129 |
+
|
| 130 |
+
def inference(image_path, vocab, label_list):
|
| 131 |
+
logger.info("building class names")
|
| 132 |
+
vocab = vocab.replace(", ", ",").replace("; ", ";")
|
| 133 |
+
demo_classes, demo_metadata = build_demo_classes_and_metadata(vocab, label_list)
|
| 134 |
+
predictor.set_metadata(demo_metadata)
|
| 135 |
+
im = cv2.imread(image_path)
|
| 136 |
+
outputs = predictor(im)
|
| 137 |
+
v = OpenVocabVisualizer(im[:, :, ::-1], demo_metadata, instance_mode=ColorMode.IMAGE)
|
| 138 |
+
panoptic_result = v.draw_panoptic_seg(outputs["panoptic_seg"][0].to("cpu"), outputs["panoptic_seg"][1]).get_image()
|
| 139 |
+
return Image.fromarray(np.uint8(panoptic_result)).convert('RGB')
|
| 140 |
+
|
| 141 |
+
|
| 142 |
+
with gr.Blocks(title=title,
|
| 143 |
+
css="""
|
| 144 |
+
#submit {background: #3498db; color: white; border: none; padding: 10px 20px; border-radius: 5px;width: 20%;margin: 0 auto; display: block;}
|
| 145 |
+
"""
|
| 146 |
+
) as demo:
|
| 147 |
+
gr.Markdown("<h1 style='text-align: center; margin-bottom: 1rem'>" + title + "</h1>")
|
| 148 |
+
input_components = []
|
| 149 |
+
output_components = []
|
| 150 |
+
|
| 151 |
+
with gr.Row():
|
| 152 |
+
output_image_gr = gr.Image(label="Panoptic Segmentation Output", type="pil")
|
| 153 |
+
output_components.append(output_image_gr)
|
| 154 |
+
|
| 155 |
+
with gr.Row().style(equal_height=True):
|
| 156 |
+
with gr.Column(scale=3, variant="panel") as input_component_column:
|
| 157 |
+
input_image_gr = gr.Image(type="filepath", label="Input Image")
|
| 158 |
+
extra_vocab_gr = gr.Textbox(label="Extra Vocabulary (separated by ;)", placeholder="house;sky")
|
| 159 |
+
category_list_gr = gr.CheckboxGroup(
|
| 160 |
+
choices=["COCO (133 categories)", "ADE (150 categories)", "LVIS (1203 categories)", "Cityscapes (19 categories)"],
|
| 161 |
+
label="Category to use",
|
| 162 |
+
)
|
| 163 |
+
input_components.extend([input_image_gr, extra_vocab_gr, category_list_gr])
|
| 164 |
+
|
| 165 |
+
with gr.Column(scale=2):
|
| 166 |
+
examples_handler = gr.Examples(
|
| 167 |
+
examples=examples,
|
| 168 |
+
inputs=[c for c in input_components if not isinstance(c, gr.State)],
|
| 169 |
+
outputs=[c for c in output_components if not isinstance(c, gr.State)],
|
| 170 |
+
fn=inference,
|
| 171 |
+
cache_examples=torch.cuda.is_available(),
|
| 172 |
+
examples_per_page=5,
|
| 173 |
+
)
|
| 174 |
+
with gr.Row():
|
| 175 |
+
clear_btn = gr.Button("Clear")
|
| 176 |
+
submit_btn = gr.Button("Submit", variant="primary")
|
| 177 |
+
|
| 178 |
+
gr.Markdown(article)
|
| 179 |
+
|
| 180 |
+
submit_btn.click(
|
| 181 |
+
inference,
|
| 182 |
+
input_components,
|
| 183 |
+
output_components,
|
| 184 |
+
api_name="predict",
|
| 185 |
+
scroll_to_output=True,
|
| 186 |
+
)
|
| 187 |
+
|
| 188 |
+
clear_btn.click(
|
| 189 |
+
None,
|
| 190 |
+
[],
|
| 191 |
+
(input_components + output_components + [input_component_column]),
|
| 192 |
+
_js=f"""() => {json.dumps(
|
| 193 |
+
[component.cleared_value if hasattr(component, "cleared_value") else None
|
| 194 |
+
for component in input_components + output_components] + (
|
| 195 |
+
[gr.Column.update(visible=True)]
|
| 196 |
+
)
|
| 197 |
+
+ ([gr.Column.update(visible=False)])
|
| 198 |
+
)}
|
| 199 |
+
""",
|
| 200 |
+
)
|
| 201 |
+
|
| 202 |
+
demo.launch(server_port=7881)
|
configs/coco/Base-COCO-PanopticSegmentation.yaml
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
MODEL:
|
| 2 |
+
BACKBONE:
|
| 3 |
+
FREEZE_AT: 0
|
| 4 |
+
NAME: "build_resnet_backbone"
|
| 5 |
+
WEIGHTS: "detectron2://ImageNetPretrained/torchvision/R-50.pkl"
|
| 6 |
+
PIXEL_MEAN: [123.675, 116.280, 103.530]
|
| 7 |
+
PIXEL_STD: [58.395, 57.120, 57.375]
|
| 8 |
+
RESNETS:
|
| 9 |
+
DEPTH: 50
|
| 10 |
+
# STEM_TYPE: "basic" # not used
|
| 11 |
+
STEM_OUT_CHANNELS: 64
|
| 12 |
+
STRIDE_IN_1X1: False
|
| 13 |
+
OUT_FEATURES: ["res2", "res3", "res4", "res5"]
|
| 14 |
+
# NORM: "SyncBN"
|
| 15 |
+
# RES5_MULTI_GRID: [1, 1, 1] # not used
|
| 16 |
+
DATASETS:
|
| 17 |
+
TRAIN: ("coco_2017_train_panoptic",)
|
| 18 |
+
TEST: ("coco_2017_val_panoptic_with_sem_seg",) # to evaluate instance and semantic performance as well
|
| 19 |
+
SOLVER:
|
| 20 |
+
IMS_PER_BATCH: 16
|
| 21 |
+
BASE_LR: 0.0001
|
| 22 |
+
STEPS: (327778, 355092)
|
| 23 |
+
MAX_ITER: 368750
|
| 24 |
+
WARMUP_FACTOR: 1.0
|
| 25 |
+
WARMUP_ITERS: 10
|
| 26 |
+
WEIGHT_DECAY: 0.05
|
| 27 |
+
OPTIMIZER: "ADAMW"
|
| 28 |
+
BACKBONE_MULTIPLIER: 0.1
|
| 29 |
+
CLIP_GRADIENTS:
|
| 30 |
+
ENABLED: True
|
| 31 |
+
CLIP_TYPE: "full_model"
|
| 32 |
+
CLIP_VALUE: 0.01
|
| 33 |
+
NORM_TYPE: 2.0
|
| 34 |
+
AMP:
|
| 35 |
+
ENABLED: True
|
| 36 |
+
INPUT:
|
| 37 |
+
IMAGE_SIZE: 1024
|
| 38 |
+
MIN_SCALE: 0.1
|
| 39 |
+
MAX_SCALE: 2.0
|
| 40 |
+
FORMAT: "RGB"
|
| 41 |
+
DATASET_MAPPER_NAME: "coco_panoptic_lsj"
|
| 42 |
+
TEST:
|
| 43 |
+
EVAL_PERIOD: 100000000
|
| 44 |
+
DATALOADER:
|
| 45 |
+
FILTER_EMPTY_ANNOTATIONS: True
|
| 46 |
+
NUM_WORKERS: 4
|
| 47 |
+
VERSION: 2
|
configs/coco/frozenseg/convnext_large_eval_a847.yaml
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
|
| 3 |
+
MODEL:
|
| 4 |
+
MASK_FORMER:
|
| 5 |
+
TEST:
|
| 6 |
+
PANOPTIC_ON: False
|
| 7 |
+
INSTANCE_ON: False
|
| 8 |
+
|
| 9 |
+
DATASETS:
|
| 10 |
+
TEST: ("openvocab_ade20k_full_sem_seg_val",)
|
configs/coco/frozenseg/convnext_large_eval_ade20k.yaml
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ../maskformer2_R50_bs16_50ep.yaml
|
| 2 |
+
MODEL:
|
| 3 |
+
META_ARCHITECTURE: "FrozenSeg"
|
| 4 |
+
SEM_SEG_HEAD:
|
| 5 |
+
NAME: "FrozenSegHead"
|
| 6 |
+
# backbone part.
|
| 7 |
+
BACKBONE:
|
| 8 |
+
NAME: "CLIP"
|
| 9 |
+
WEIGHTS: ""
|
| 10 |
+
PIXEL_MEAN: [122.7709383, 116.7460125, 104.09373615]
|
| 11 |
+
PIXEL_STD: [68.5005327, 66.6321579, 70.32316305]
|
| 12 |
+
FROZEN_SEG:
|
| 13 |
+
CLIP_MODEL_NAME: "convnext_large_d_320"
|
| 14 |
+
# CLIP_PRETRAINED_WEIGHTS: "laion2b_s29b_b131k_ft_soup"
|
| 15 |
+
CLIP_PRETRAINED_WEIGHTS: "pretrained_checkpoint/models--laion--CLIP-convnext_large_d_320.laion2B-s29B-b131K-ft-soup/open_clip_pytorch_model.bin"
|
| 16 |
+
EMBED_DIM: 768
|
| 17 |
+
GEOMETRIC_ENSEMBLE_ALPHA: 0.4
|
| 18 |
+
GEOMETRIC_ENSEMBLE_BETA: 0.8
|
| 19 |
+
MASK_FORMER:
|
| 20 |
+
NUM_OBJECT_QUERIES: 250
|
| 21 |
+
TEST:
|
| 22 |
+
SEMANTIC_ON: True
|
| 23 |
+
INSTANCE_ON: True
|
| 24 |
+
PANOPTIC_ON: True
|
| 25 |
+
OBJECT_MASK_THRESHOLD: 0.0
|
| 26 |
+
|
| 27 |
+
DATASETS:
|
| 28 |
+
TRAIN: ("openvocab_coco_2017_train_panoptic_with_sem_seg",)
|
| 29 |
+
TEST: ("openvocab_ade20k_panoptic_val",)
|
configs/coco/frozenseg/convnext_large_eval_bdd_panop.yaml
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
|
| 3 |
+
MODEL:
|
| 4 |
+
MASK_FORMER:
|
| 5 |
+
TEST:
|
| 6 |
+
PANOPTIC_ON: True
|
| 7 |
+
INSTANCE_ON: False
|
| 8 |
+
SEMANTIC_ON: False
|
| 9 |
+
OBJECT_MASK_THRESHOLD: 0.4
|
| 10 |
+
INPUT:
|
| 11 |
+
MIN_SIZE_TEST: 800
|
| 12 |
+
MAX_SIZE_TEST: 1333
|
| 13 |
+
DATASETS:
|
| 14 |
+
TEST: ("bdd10k_40_panoptic_val",)
|
configs/coco/frozenseg/convnext_large_eval_bdd_sem.yaml
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
|
| 3 |
+
MODEL:
|
| 4 |
+
MASK_FORMER:
|
| 5 |
+
TEST:
|
| 6 |
+
PANOPTIC_ON: False
|
| 7 |
+
INSTANCE_ON: False
|
| 8 |
+
SEMANTIC_ON: True
|
| 9 |
+
INPUT:
|
| 10 |
+
MIN_SIZE_TEST: 800
|
| 11 |
+
MAX_SIZE_TEST: 1333
|
| 12 |
+
DATASETS:
|
| 13 |
+
TEST: ("bdd10k_val_sem_seg",)
|
configs/coco/frozenseg/convnext_large_eval_cityscapes.yaml
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
|
| 3 |
+
INPUT:
|
| 4 |
+
MIN_SIZE_TEST: 1024
|
| 5 |
+
MAX_SIZE_TEST: 2560
|
| 6 |
+
|
| 7 |
+
DATASETS:
|
| 8 |
+
TEST: ("openvocab_cityscapes_fine_panoptic_val",)
|
configs/coco/frozenseg/convnext_large_eval_coco.yaml
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
DATASETS:
|
| 3 |
+
TEST: ("openvocab_coco_2017_val_panoptic_with_sem_seg",)
|
configs/coco/frozenseg/convnext_large_eval_lvis.yaml
ADDED
|
@@ -0,0 +1,11 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
|
| 3 |
+
MODEL:
|
| 4 |
+
MASK_FORMER:
|
| 5 |
+
TEST:
|
| 6 |
+
PANOPTIC_ON: False
|
| 7 |
+
INSTANCE_ON: True
|
| 8 |
+
SEMANTIC_ON: False
|
| 9 |
+
|
| 10 |
+
DATASETS:
|
| 11 |
+
TEST: ("openvocab_lvis_v1_val",)
|
configs/coco/frozenseg/convnext_large_eval_mapillary_vistas.yaml
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
|
| 3 |
+
MODEL:
|
| 4 |
+
MASK_FORMER:
|
| 5 |
+
TEST:
|
| 6 |
+
INSTANCE_ON: False
|
| 7 |
+
INPUT:
|
| 8 |
+
MIN_SIZE_TEST: 1024
|
| 9 |
+
MAX_SIZE_TEST: 2560
|
| 10 |
+
|
| 11 |
+
DATASETS:
|
| 12 |
+
TEST: ("openvocab_mapillary_vistas_panoptic_val",)
|
configs/coco/frozenseg/convnext_large_eval_pas21.yaml
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
|
| 3 |
+
MODEL:
|
| 4 |
+
MASK_FORMER:
|
| 5 |
+
TEST:
|
| 6 |
+
PANOPTIC_ON: False
|
| 7 |
+
INSTANCE_ON: False
|
| 8 |
+
|
| 9 |
+
DATASETS:
|
| 10 |
+
TEST: ("openvocab_pascal21_sem_seg_val",)
|
configs/coco/frozenseg/convnext_large_eval_pc459.yaml
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
|
| 3 |
+
MODEL:
|
| 4 |
+
MASK_FORMER:
|
| 5 |
+
TEST:
|
| 6 |
+
PANOPTIC_ON: False
|
| 7 |
+
INSTANCE_ON: False
|
| 8 |
+
|
| 9 |
+
DATASETS:
|
| 10 |
+
TEST: ("openvocab_pascal_ctx459_sem_seg_val",)
|
configs/coco/frozenseg/r50x64_eval_ade20k.yaml
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: ./convnext_large_eval_ade20k.yaml
|
| 2 |
+
MODEL:
|
| 3 |
+
FROZEN_SEG:
|
| 4 |
+
CLIP_MODEL_NAME: "RN50x64"
|
| 5 |
+
CLIP_PRETRAINED_WEIGHTS: "openai"
|
| 6 |
+
EMBED_DIM: 1024
|
| 7 |
+
ENSEMBLE_ON_VALID_MASK: True
|
| 8 |
+
MASK_FORMER:
|
| 9 |
+
TEST:
|
| 10 |
+
PANOPTIC_ON: False
|
| 11 |
+
INSTANCE_ON: False
|
| 12 |
+
DATASETS:
|
| 13 |
+
TEST: ("openvocab_ade20k_full_sem_seg_val",)
|
configs/coco/maskformer2_R50_bs16_50ep.yaml
ADDED
|
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
_BASE_: Base-COCO-PanopticSegmentation.yaml
|
| 2 |
+
MODEL:
|
| 3 |
+
META_ARCHITECTURE: "MaskFormer"
|
| 4 |
+
SEM_SEG_HEAD:
|
| 5 |
+
NAME: "MaskFormerHead"
|
| 6 |
+
IN_FEATURES: ["res2", "res3", "res4", "res5"]
|
| 7 |
+
IGNORE_VALUE: 255
|
| 8 |
+
NUM_CLASSES: 133
|
| 9 |
+
LOSS_WEIGHT: 1.0
|
| 10 |
+
CONVS_DIM: 256
|
| 11 |
+
MASK_DIM: 256
|
| 12 |
+
NORM: "GN"
|
| 13 |
+
# pixel decoder
|
| 14 |
+
PIXEL_DECODER_NAME: "MSDeformAttnPixelDecoder"
|
| 15 |
+
IN_FEATURES: ["res2", "res3", "res4", "res5"]
|
| 16 |
+
DEFORMABLE_TRANSFORMER_ENCODER_IN_FEATURES: ["res3", "res4", "res5"]
|
| 17 |
+
COMMON_STRIDE: 4
|
| 18 |
+
TRANSFORMER_ENC_LAYERS: 6
|
| 19 |
+
MASK_FORMER:
|
| 20 |
+
TRANSFORMER_DECODER_NAME: "MultiScaleMaskedTransformerDecoder"
|
| 21 |
+
TRANSFORMER_IN_FEATURE: "multi_scale_pixel_decoder"
|
| 22 |
+
DEEP_SUPERVISION: True
|
| 23 |
+
NO_OBJECT_WEIGHT: 0.1
|
| 24 |
+
CLASS_WEIGHT: 2.0
|
| 25 |
+
MASK_WEIGHT: 5.0
|
| 26 |
+
DICE_WEIGHT: 5.0
|
| 27 |
+
HIDDEN_DIM: 256
|
| 28 |
+
NUM_OBJECT_QUERIES: 100
|
| 29 |
+
NHEADS: 8
|
| 30 |
+
DROPOUT: 0.0
|
| 31 |
+
DIM_FEEDFORWARD: 2048
|
| 32 |
+
ENC_LAYERS: 0
|
| 33 |
+
PRE_NORM: False
|
| 34 |
+
ENFORCE_INPUT_PROJ: False
|
| 35 |
+
SIZE_DIVISIBILITY: 32
|
| 36 |
+
DEC_LAYERS: 10 # 9 decoder layers, add one for the loss on learnable query
|
| 37 |
+
TRAIN_NUM_POINTS: 12544
|
| 38 |
+
OVERSAMPLE_RATIO: 3.0
|
| 39 |
+
IMPORTANCE_SAMPLE_RATIO: 0.75
|
| 40 |
+
TEST:
|
| 41 |
+
SEMANTIC_ON: True
|
| 42 |
+
INSTANCE_ON: True
|
| 43 |
+
PANOPTIC_ON: True
|
| 44 |
+
OVERLAP_THRESHOLD: 0.8
|
| 45 |
+
OBJECT_MASK_THRESHOLD: 0.8
|
datasets/README.md
ADDED
|
@@ -0,0 +1,262 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Prepare Datasets for FrozenSeg
|
| 2 |
+
|
| 3 |
+
A dataset can be used by accessing [DatasetCatalog](https://detectron2.readthedocs.io/modules/data.html#detectron2.data.DatasetCatalog)
|
| 4 |
+
for its data, or [MetadataCatalog](https://detectron2.readthedocs.io/modules/data.html#detectron2.data.MetadataCatalog) for its metadata (class names, etc).
|
| 5 |
+
This document explains how to setup the builtin datasets so they can be used by the above APIs.
|
| 6 |
+
[Use Custom Datasets](https://detectron2.readthedocs.io/tutorials/datasets.html) gives a deeper dive on how to use `DatasetCatalog` and `MetadataCatalog`,
|
| 7 |
+
and how to add new datasets to them.
|
| 8 |
+
|
| 9 |
+
FrozenSeg has builtin support for a few datasets.
|
| 10 |
+
The datasets are assumed to exist in a directory specified by the environment variable
|
| 11 |
+
`DETECTRON2_DATASETS`.
|
| 12 |
+
Under this directory, detectron2 will look for datasets in the structure described below, if needed.
|
| 13 |
+
```
|
| 14 |
+
$DETECTRON2_DATASETS/
|
| 15 |
+
# panoptic datasets
|
| 16 |
+
ADEChallengeData2016/
|
| 17 |
+
coco/
|
| 18 |
+
cityscapes/
|
| 19 |
+
mapillary_vistas/
|
| 20 |
+
bdd100k/
|
| 21 |
+
# semantic datasets
|
| 22 |
+
VOCdevkit/
|
| 23 |
+
ADE20K_2021_17_01/
|
| 24 |
+
pascal_ctx_d2/
|
| 25 |
+
pascal_voc_d2/
|
| 26 |
+
```
|
| 27 |
+
|
| 28 |
+
You can set the location for builtin datasets by `export DETECTRON2_DATASETS=/path/to/datasets`.
|
| 29 |
+
If left unset, the default is `./datasets` relative to your current working directory.
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
## Expected dataset structure for [COCO](https://cocodataset.org/#download):
|
| 33 |
+
|
| 34 |
+
```
|
| 35 |
+
coco/
|
| 36 |
+
annotations/
|
| 37 |
+
instances_{train,val}2017.json
|
| 38 |
+
panoptic_{train,val}2017.json
|
| 39 |
+
{train,val}2017/
|
| 40 |
+
# image files that are mentioned in the corresponding json
|
| 41 |
+
panoptic_{train,val}2017/ # png annotations
|
| 42 |
+
panoptic_semseg_{train,val}2017/ # generated by the script mentioned below
|
| 43 |
+
```
|
| 44 |
+
|
| 45 |
+
Install panopticapi by:
|
| 46 |
+
```
|
| 47 |
+
pip install git+https://github.com/cocodataset/panopticapi.git
|
| 48 |
+
```
|
| 49 |
+
Then, run `python datasets/prepare_coco_semantic_annos_from_panoptic_annos.py`, to extract semantic annotations from panoptic annotations (only used for evaluation).
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
## Expected dataset structure for [cityscapes](https://www.cityscapes-dataset.com/downloads/):
|
| 53 |
+
```
|
| 54 |
+
cityscapes/
|
| 55 |
+
gtFine/
|
| 56 |
+
train/
|
| 57 |
+
aachen/
|
| 58 |
+
color.png, instanceIds.png, labelIds.png, polygons.json,
|
| 59 |
+
labelTrainIds.png
|
| 60 |
+
...
|
| 61 |
+
val/
|
| 62 |
+
test/
|
| 63 |
+
# below are generated Cityscapes panoptic annotation
|
| 64 |
+
cityscapes_panoptic_train.json
|
| 65 |
+
cityscapes_panoptic_train/
|
| 66 |
+
cityscapes_panoptic_val.json
|
| 67 |
+
cityscapes_panoptic_val/
|
| 68 |
+
cityscapes_panoptic_test.json
|
| 69 |
+
cityscapes_panoptic_test/
|
| 70 |
+
leftImg8bit/
|
| 71 |
+
train/
|
| 72 |
+
val/
|
| 73 |
+
test/
|
| 74 |
+
```
|
| 75 |
+
Install cityscapes scripts by:
|
| 76 |
+
```
|
| 77 |
+
pip install git+https://github.com/mcordts/cityscapesScripts.git
|
| 78 |
+
```
|
| 79 |
+
|
| 80 |
+
Note: to create labelTrainIds.png, first prepare the above structure, then run cityscapesescript with:
|
| 81 |
+
```
|
| 82 |
+
CITYSCAPES_DATASET=/path/to/abovementioned/cityscapes python cityscapesscripts/preparation/createTrainIdLabelImgs.py
|
| 83 |
+
```
|
| 84 |
+
These files are not needed for instance segmentation.
|
| 85 |
+
|
| 86 |
+
Note: to generate Cityscapes panoptic dataset, run cityscapesescript with:
|
| 87 |
+
```
|
| 88 |
+
CITYSCAPES_DATASET=/path/to/abovementioned/cityscapes python cityscapesscripts/preparation/createPanopticImgs.py
|
| 89 |
+
```
|
| 90 |
+
These files are not needed for semantic and instance segmentation.
|
| 91 |
+
|
| 92 |
+
|
| 93 |
+
## Expected dataset structure for [ADE20k (A150)](http://sceneparsing.csail.mit.edu/):
|
| 94 |
+
```
|
| 95 |
+
ADEChallengeData2016/
|
| 96 |
+
images/
|
| 97 |
+
annotations/
|
| 98 |
+
objectInfo150.txt
|
| 99 |
+
# download instance annotation
|
| 100 |
+
annotations_instance/
|
| 101 |
+
# generated by prepare_ade20k_sem_seg.py
|
| 102 |
+
annotations_detectron2/
|
| 103 |
+
# below are generated by prepare_ade20k_pan_seg.py
|
| 104 |
+
ade20k_panoptic_{train,val}.json
|
| 105 |
+
ade20k_panoptic_{train,val}/
|
| 106 |
+
# below are generated by prepare_ade20k_ins_seg.py
|
| 107 |
+
ade20k_instance_{train,val}.json
|
| 108 |
+
```
|
| 109 |
+
|
| 110 |
+
The directory `annotations_detectron2` is generated by running `python datasets/prepare_ade20k_sem_seg.py`.
|
| 111 |
+
|
| 112 |
+
Install panopticapi by:
|
| 113 |
+
```bash
|
| 114 |
+
pip install git+https://github.com/cocodataset/panopticapi.git
|
| 115 |
+
```
|
| 116 |
+
|
| 117 |
+
Download the instance annotation from http://sceneparsing.csail.mit.edu/:
|
| 118 |
+
```bash
|
| 119 |
+
wget http://sceneparsing.csail.mit.edu/data/ChallengeData2017/annotations_instance.tar
|
| 120 |
+
```
|
| 121 |
+
|
| 122 |
+
Then, run `python datasets/prepare_ade20k_pan_seg.py`, to combine semantic and instance annotations for panoptic annotations.
|
| 123 |
+
|
| 124 |
+
And run `python datasets/prepare_ade20k_ins_seg.py`, to extract instance annotations in COCO format.
|
| 125 |
+
|
| 126 |
+
|
| 127 |
+
## Expected dataset structure for [Mapillary Vistas](https://www.mapillary.com/dataset/vistas):
|
| 128 |
+
```
|
| 129 |
+
mapillary_vistas/
|
| 130 |
+
training/
|
| 131 |
+
images/
|
| 132 |
+
instances/
|
| 133 |
+
labels/
|
| 134 |
+
panoptic/
|
| 135 |
+
validation/
|
| 136 |
+
images/
|
| 137 |
+
instances/
|
| 138 |
+
labels/
|
| 139 |
+
panoptic/
|
| 140 |
+
```
|
| 141 |
+
|
| 142 |
+
No preprocessing is needed for Mapillary Vistas on semantic and panoptic segmentation.
|
| 143 |
+
|
| 144 |
+
## Expected dataset structure for [BDD100K](https://doc.bdd100k.com/download.html#id1)
|
| 145 |
+
```
|
| 146 |
+
bdd100k/
|
| 147 |
+
images/
|
| 148 |
+
10k/
|
| 149 |
+
train/
|
| 150 |
+
val/
|
| 151 |
+
test/
|
| 152 |
+
json
|
| 153 |
+
labels/
|
| 154 |
+
pan_seg/
|
| 155 |
+
sem_seg/
|
| 156 |
+
```
|
| 157 |
+
|
| 158 |
+
`coco-format` annotations is obtained by running:
|
| 159 |
+
|
| 160 |
+
|
| 161 |
+
```
|
| 162 |
+
cd $DETECTRON2_DATASETS
|
| 163 |
+
wget https://github.com/chenxi52/FrozenSeg/releases/download/latest/bdd100k_json.zip
|
| 164 |
+
unzip bdd100k_json.zip
|
| 165 |
+
```
|
| 166 |
+
|
| 167 |
+
|
| 168 |
+
## Expected dataset structure for [ADE20k-Full (A-847)](https://groups.csail.mit.edu/vision/datasets/ADE20K/):
|
| 169 |
+
```
|
| 170 |
+
ADE20K_2021_17_01/
|
| 171 |
+
images/
|
| 172 |
+
index_ade20k.pkl
|
| 173 |
+
objects.txt
|
| 174 |
+
# generated by prepare_ade20k_full_sem_seg.py
|
| 175 |
+
images_detectron2/
|
| 176 |
+
annotations_detectron2/
|
| 177 |
+
```
|
| 178 |
+
|
| 179 |
+
Register and download the dataset from https://groups.csail.mit.edu/vision/datasets/ADE20K/:
|
| 180 |
+
```bash
|
| 181 |
+
cd $DETECTRON2_DATASETS
|
| 182 |
+
wget your/personal/download/link/{username}_{hash}.zip
|
| 183 |
+
unzip {username}_{hash}.zip
|
| 184 |
+
```
|
| 185 |
+
|
| 186 |
+
Generate the directories `ADE20K_2021_17_01/images_detectron2` and `ADE20K_2021_17_01/annotations_detectron2` by running:
|
| 187 |
+
```bash
|
| 188 |
+
python datasets/prepare_ade20k_full_sem_seg.py
|
| 189 |
+
```
|
| 190 |
+
|
| 191 |
+
## Expected dataset structure for [PASCAL Context Full (PC-459)](https://www.cs.stanford.edu/~roozbeh/pascal-context/) and [PASCAL VOC (PAS-21)](http://host.robots.ox.ac.uk/pascal/VOC/):
|
| 192 |
+
|
| 193 |
+
```bash
|
| 194 |
+
VOCdevkit/
|
| 195 |
+
VOC2012/
|
| 196 |
+
Annotations/
|
| 197 |
+
JPEGImages/
|
| 198 |
+
ImageSets/
|
| 199 |
+
Segmentation/
|
| 200 |
+
VOC2010/
|
| 201 |
+
JPEGImages/
|
| 202 |
+
trainval/
|
| 203 |
+
trainval_merged.json
|
| 204 |
+
# generated by prepare_pascal_voc_sem_seg.py
|
| 205 |
+
pascal_voc_d2/
|
| 206 |
+
images/
|
| 207 |
+
annotations_pascal21/
|
| 208 |
+
# pascal 20 excludes the background class
|
| 209 |
+
annotations_pascal20/
|
| 210 |
+
# generated by prepare_pascal_ctx_sem_seg.py
|
| 211 |
+
pascal_ctx_d2/
|
| 212 |
+
images/
|
| 213 |
+
annotations_ctx59/
|
| 214 |
+
# generated by prepare_pascal_ctx_full_sem_seg.py
|
| 215 |
+
annotations_ctx459/
|
| 216 |
+
|
| 217 |
+
```
|
| 218 |
+
### PASCAL VOC (PAS-21)
|
| 219 |
+
|
| 220 |
+
Download the dataset from http://host.robots.ox.ac.uk/pascal/VOC/:
|
| 221 |
+
```bash
|
| 222 |
+
cd $DETECTRON2_DATASETS
|
| 223 |
+
wget http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar
|
| 224 |
+
# generate folder VOCdevkit/VOC2012
|
| 225 |
+
tar -xvf VOCtrainval_11-May-2012.tar
|
| 226 |
+
```
|
| 227 |
+
|
| 228 |
+
Generate directory `pascal_voc_d2` running:
|
| 229 |
+
```bash
|
| 230 |
+
python datasets/prepare_pascal_voc_sem_seg.py
|
| 231 |
+
```
|
| 232 |
+
|
| 233 |
+
### PASCAL Context Full (PC-459)
|
| 234 |
+
|
| 235 |
+
Download the dataset from http://host.robots.ox.ac.uk/pascal/VOC/ and annotation from https://www.cs.stanford.edu/~roozbeh/pascal-context/:
|
| 236 |
+
```bash
|
| 237 |
+
cd $DETECTRON2_DATASETS
|
| 238 |
+
wget http://host.robots.ox.ac.uk/pascal/VOC/voc2010/VOCtrainval_03-May-2010.tar
|
| 239 |
+
# generate folder VOCdevkit/VOC2010
|
| 240 |
+
tar -xvf VOCtrainval_03-May-2010.tar
|
| 241 |
+
wget https://www.cs.stanford.edu/~roozbeh/pascal-context/trainval.tar.gz
|
| 242 |
+
# generate folder VOCdevkit/VOC2010/trainval
|
| 243 |
+
tar -xvzf trainval.tar.gz -C VOCdevkit/VOC2010
|
| 244 |
+
wget https://codalabuser.blob.core.windows.net/public/trainval_merged.json -P VOCdevkit/VOC2010/
|
| 245 |
+
```
|
| 246 |
+
|
| 247 |
+
Install [Detail API](https://github.com/zhanghang1989/detail-api) by:
|
| 248 |
+
```bash
|
| 249 |
+
git clone https://github.com/zhanghang1989/detail-api.git
|
| 250 |
+
rm detail-api/PythonAPI/detail/_mask.c
|
| 251 |
+
pip install -e detail-api/PythonAPI/
|
| 252 |
+
```
|
| 253 |
+
|
| 254 |
+
Generate directory `pascal_ctx_d2/images` running:
|
| 255 |
+
```bash
|
| 256 |
+
python datasets/prepare_pascal_ctx_sem_seg.py
|
| 257 |
+
```
|
| 258 |
+
|
| 259 |
+
Generate directory `pascal_ctx_d2/annotations_ctx459` running:
|
| 260 |
+
```bash
|
| 261 |
+
python datasets/prepare_pascal_ctx_full_sem_seg.py
|
| 262 |
+
```
|
datasets/ade20k_instance_catid_mapping.txt
ADDED
|
@@ -0,0 +1,104 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Instacne100 SceneParse150 FullADE20K
|
| 2 |
+
1 8 165
|
| 3 |
+
2 9 3055
|
| 4 |
+
3 11 350
|
| 5 |
+
4 13 1831
|
| 6 |
+
5 15 774
|
| 7 |
+
5 15 783
|
| 8 |
+
6 16 2684
|
| 9 |
+
7 19 687
|
| 10 |
+
8 20 471
|
| 11 |
+
9 21 401
|
| 12 |
+
10 23 1735
|
| 13 |
+
11 24 2473
|
| 14 |
+
12 25 2329
|
| 15 |
+
13 28 1564
|
| 16 |
+
14 31 57
|
| 17 |
+
15 32 2272
|
| 18 |
+
16 33 907
|
| 19 |
+
17 34 724
|
| 20 |
+
18 36 2985
|
| 21 |
+
18 36 533
|
| 22 |
+
19 37 1395
|
| 23 |
+
20 38 155
|
| 24 |
+
21 39 2053
|
| 25 |
+
22 40 689
|
| 26 |
+
23 42 266
|
| 27 |
+
24 43 581
|
| 28 |
+
25 44 2380
|
| 29 |
+
26 45 491
|
| 30 |
+
27 46 627
|
| 31 |
+
28 48 2388
|
| 32 |
+
29 50 943
|
| 33 |
+
30 51 2096
|
| 34 |
+
31 54 2530
|
| 35 |
+
32 56 420
|
| 36 |
+
33 57 1948
|
| 37 |
+
34 58 1869
|
| 38 |
+
35 59 2251
|
| 39 |
+
36 63 239
|
| 40 |
+
37 65 571
|
| 41 |
+
38 66 2793
|
| 42 |
+
39 67 978
|
| 43 |
+
40 68 236
|
| 44 |
+
41 70 181
|
| 45 |
+
42 71 629
|
| 46 |
+
43 72 2598
|
| 47 |
+
44 73 1744
|
| 48 |
+
45 74 1374
|
| 49 |
+
46 75 591
|
| 50 |
+
47 76 2679
|
| 51 |
+
48 77 223
|
| 52 |
+
49 79 47
|
| 53 |
+
50 81 327
|
| 54 |
+
51 82 2821
|
| 55 |
+
52 83 1451
|
| 56 |
+
53 84 2880
|
| 57 |
+
54 86 480
|
| 58 |
+
55 87 77
|
| 59 |
+
56 88 2616
|
| 60 |
+
57 89 246
|
| 61 |
+
57 89 247
|
| 62 |
+
58 90 2733
|
| 63 |
+
59 91 14
|
| 64 |
+
60 93 38
|
| 65 |
+
61 94 1936
|
| 66 |
+
62 96 120
|
| 67 |
+
63 98 1702
|
| 68 |
+
64 99 249
|
| 69 |
+
65 103 2928
|
| 70 |
+
66 104 2337
|
| 71 |
+
67 105 1023
|
| 72 |
+
68 108 2989
|
| 73 |
+
69 109 1930
|
| 74 |
+
70 111 2586
|
| 75 |
+
71 112 131
|
| 76 |
+
72 113 146
|
| 77 |
+
73 116 95
|
| 78 |
+
74 117 1563
|
| 79 |
+
75 119 1708
|
| 80 |
+
76 120 103
|
| 81 |
+
77 121 1002
|
| 82 |
+
78 122 2569
|
| 83 |
+
79 124 2833
|
| 84 |
+
80 125 1551
|
| 85 |
+
81 126 1981
|
| 86 |
+
82 127 29
|
| 87 |
+
83 128 187
|
| 88 |
+
84 130 747
|
| 89 |
+
85 131 2254
|
| 90 |
+
86 133 2262
|
| 91 |
+
87 134 1260
|
| 92 |
+
88 135 2243
|
| 93 |
+
89 136 2932
|
| 94 |
+
90 137 2836
|
| 95 |
+
91 138 2850
|
| 96 |
+
92 139 64
|
| 97 |
+
93 140 894
|
| 98 |
+
94 143 1919
|
| 99 |
+
95 144 1583
|
| 100 |
+
96 145 318
|
| 101 |
+
97 147 2046
|
| 102 |
+
98 148 1098
|
| 103 |
+
99 149 530
|
| 104 |
+
100 150 954
|
datasets/ade20k_instance_imgCatIds.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
datasets/prepare_ade20k_full_sem_seg.py
ADDED
|
@@ -0,0 +1,1004 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import pickle as pkl
|
| 3 |
+
from pathlib import Path
|
| 4 |
+
|
| 5 |
+
import cv2
|
| 6 |
+
import numpy as np
|
| 7 |
+
import tqdm
|
| 8 |
+
from PIL import Image
|
| 9 |
+
|
| 10 |
+
ADE20K_SEM_SEG_FULL_CATEGORIES = [
|
| 11 |
+
{"name": "wall", "id": 2978, "trainId": 0},
|
| 12 |
+
{"name": "building, edifice", "id": 312, "trainId": 1},
|
| 13 |
+
{"name": "sky", "id": 2420, "trainId": 2},
|
| 14 |
+
{"name": "tree", "id": 2855, "trainId": 3},
|
| 15 |
+
{"name": "road, route", "id": 2131, "trainId": 4},
|
| 16 |
+
{"name": "floor, flooring", "id": 976, "trainId": 5},
|
| 17 |
+
{"name": "ceiling", "id": 447, "trainId": 6},
|
| 18 |
+
{"name": "bed", "id": 165, "trainId": 7},
|
| 19 |
+
{"name": "sidewalk, pavement", "id": 2377, "trainId": 8},
|
| 20 |
+
{"name": "earth, ground", "id": 838, "trainId": 9},
|
| 21 |
+
{"name": "cabinet", "id": 350, "trainId": 10},
|
| 22 |
+
{"name": "person, individual, someone, somebody, mortal, soul", "id": 1831, "trainId": 11},
|
| 23 |
+
{"name": "grass", "id": 1125, "trainId": 12},
|
| 24 |
+
{"name": "windowpane, window", "id": 3055, "trainId": 13},
|
| 25 |
+
{"name": "car, auto, automobile, machine, motorcar", "id": 401, "trainId": 14},
|
| 26 |
+
{"name": "mountain, mount", "id": 1610, "trainId": 15},
|
| 27 |
+
{"name": "plant, flora, plant life", "id": 1910, "trainId": 16},
|
| 28 |
+
{"name": "table", "id": 2684, "trainId": 17},
|
| 29 |
+
{"name": "chair", "id": 471, "trainId": 18},
|
| 30 |
+
{"name": "curtain, drape, drapery, mantle, pall", "id": 687, "trainId": 19},
|
| 31 |
+
{"name": "door", "id": 774, "trainId": 20},
|
| 32 |
+
{"name": "sofa, couch, lounge", "id": 2473, "trainId": 21},
|
| 33 |
+
{"name": "sea", "id": 2264, "trainId": 22},
|
| 34 |
+
{"name": "painting, picture", "id": 1735, "trainId": 23},
|
| 35 |
+
{"name": "water", "id": 2994, "trainId": 24},
|
| 36 |
+
{"name": "mirror", "id": 1564, "trainId": 25},
|
| 37 |
+
{"name": "house", "id": 1276, "trainId": 26},
|
| 38 |
+
{"name": "rug, carpet, carpeting", "id": 2178, "trainId": 27},
|
| 39 |
+
{"name": "shelf", "id": 2329, "trainId": 28},
|
| 40 |
+
{"name": "armchair", "id": 57, "trainId": 29},
|
| 41 |
+
{"name": "fence, fencing", "id": 907, "trainId": 30},
|
| 42 |
+
{"name": "field", "id": 913, "trainId": 31},
|
| 43 |
+
{"name": "lamp", "id": 1395, "trainId": 32},
|
| 44 |
+
{"name": "rock, stone", "id": 2138, "trainId": 33},
|
| 45 |
+
{"name": "seat", "id": 2272, "trainId": 34},
|
| 46 |
+
{"name": "river", "id": 2128, "trainId": 35},
|
| 47 |
+
{"name": "desk", "id": 724, "trainId": 36},
|
| 48 |
+
{"name": "bathtub, bathing tub, bath, tub", "id": 155, "trainId": 37},
|
| 49 |
+
{"name": "railing, rail", "id": 2053, "trainId": 38},
|
| 50 |
+
{"name": "signboard, sign", "id": 2380, "trainId": 39},
|
| 51 |
+
{"name": "cushion", "id": 689, "trainId": 40},
|
| 52 |
+
{"name": "path", "id": 1788, "trainId": 41},
|
| 53 |
+
{"name": "work surface", "id": 3087, "trainId": 42},
|
| 54 |
+
{"name": "stairs, steps", "id": 2530, "trainId": 43},
|
| 55 |
+
{"name": "column, pillar", "id": 581, "trainId": 44},
|
| 56 |
+
{"name": "sink", "id": 2388, "trainId": 45},
|
| 57 |
+
{"name": "wardrobe, closet, press", "id": 2985, "trainId": 46},
|
| 58 |
+
{"name": "snow", "id": 2454, "trainId": 47},
|
| 59 |
+
{"name": "refrigerator, icebox", "id": 2096, "trainId": 48},
|
| 60 |
+
{"name": "base, pedestal, stand", "id": 137, "trainId": 49},
|
| 61 |
+
{"name": "bridge, span", "id": 294, "trainId": 50},
|
| 62 |
+
{"name": "blind, screen", "id": 212, "trainId": 51},
|
| 63 |
+
{"name": "runway", "id": 2185, "trainId": 52},
|
| 64 |
+
{"name": "cliff, drop, drop-off", "id": 524, "trainId": 53},
|
| 65 |
+
{"name": "sand", "id": 2212, "trainId": 54},
|
| 66 |
+
{"name": "fireplace, hearth, open fireplace", "id": 943, "trainId": 55},
|
| 67 |
+
{"name": "pillow", "id": 1869, "trainId": 56},
|
| 68 |
+
{"name": "screen door, screen", "id": 2251, "trainId": 57},
|
| 69 |
+
{"name": "toilet, can, commode, crapper, pot, potty, stool, throne", "id": 2793, "trainId": 58},
|
| 70 |
+
{"name": "skyscraper", "id": 2423, "trainId": 59},
|
| 71 |
+
{"name": "grandstand, covered stand", "id": 1121, "trainId": 60},
|
| 72 |
+
{"name": "box", "id": 266, "trainId": 61},
|
| 73 |
+
{"name": "pool table, billiard table, snooker table", "id": 1948, "trainId": 62},
|
| 74 |
+
{"name": "palm, palm tree", "id": 1744, "trainId": 63},
|
| 75 |
+
{"name": "double door", "id": 783, "trainId": 64},
|
| 76 |
+
{"name": "coffee table, cocktail table", "id": 571, "trainId": 65},
|
| 77 |
+
{"name": "counter", "id": 627, "trainId": 66},
|
| 78 |
+
{"name": "countertop", "id": 629, "trainId": 67},
|
| 79 |
+
{"name": "chest of drawers, chest, bureau, dresser", "id": 491, "trainId": 68},
|
| 80 |
+
{"name": "kitchen island", "id": 1374, "trainId": 69},
|
| 81 |
+
{"name": "boat", "id": 223, "trainId": 70},
|
| 82 |
+
{"name": "waterfall, falls", "id": 3016, "trainId": 71},
|
| 83 |
+
{
|
| 84 |
+
"name": "stove, kitchen stove, range, kitchen range, cooking stove",
|
| 85 |
+
"id": 2598,
|
| 86 |
+
"trainId": 72,
|
| 87 |
+
},
|
| 88 |
+
{"name": "flower", "id": 978, "trainId": 73},
|
| 89 |
+
{"name": "bookcase", "id": 239, "trainId": 74},
|
| 90 |
+
{"name": "controls", "id": 608, "trainId": 75},
|
| 91 |
+
{"name": "book", "id": 236, "trainId": 76},
|
| 92 |
+
{"name": "stairway, staircase", "id": 2531, "trainId": 77},
|
| 93 |
+
{"name": "streetlight, street lamp", "id": 2616, "trainId": 78},
|
| 94 |
+
{
|
| 95 |
+
"name": "computer, computing machine, computing device, data processor, electronic computer, information processing system",
|
| 96 |
+
"id": 591,
|
| 97 |
+
"trainId": 79,
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"name": "bus, autobus, coach, charabanc, double-decker, jitney, motorbus, motorcoach, omnibus, passenger vehicle",
|
| 101 |
+
"id": 327,
|
| 102 |
+
"trainId": 80,
|
| 103 |
+
},
|
| 104 |
+
{"name": "swivel chair", "id": 2679, "trainId": 81},
|
| 105 |
+
{"name": "light, light source", "id": 1451, "trainId": 82},
|
| 106 |
+
{"name": "bench", "id": 181, "trainId": 83},
|
| 107 |
+
{"name": "case, display case, showcase, vitrine", "id": 420, "trainId": 84},
|
| 108 |
+
{"name": "towel", "id": 2821, "trainId": 85},
|
| 109 |
+
{"name": "fountain", "id": 1023, "trainId": 86},
|
| 110 |
+
{"name": "embankment", "id": 855, "trainId": 87},
|
| 111 |
+
{
|
| 112 |
+
"name": "television receiver, television, television set, tv, tv set, idiot box, boob tube, telly, goggle box",
|
| 113 |
+
"id": 2733,
|
| 114 |
+
"trainId": 88,
|
| 115 |
+
},
|
| 116 |
+
{"name": "van", "id": 2928, "trainId": 89},
|
| 117 |
+
{"name": "hill", "id": 1240, "trainId": 90},
|
| 118 |
+
{"name": "awning, sunshade, sunblind", "id": 77, "trainId": 91},
|
| 119 |
+
{"name": "poster, posting, placard, notice, bill, card", "id": 1969, "trainId": 92},
|
| 120 |
+
{"name": "truck, motortruck", "id": 2880, "trainId": 93},
|
| 121 |
+
{"name": "airplane, aeroplane, plane", "id": 14, "trainId": 94},
|
| 122 |
+
{"name": "pole", "id": 1936, "trainId": 95},
|
| 123 |
+
{"name": "tower", "id": 2828, "trainId": 96},
|
| 124 |
+
{"name": "court", "id": 631, "trainId": 97},
|
| 125 |
+
{"name": "ball", "id": 103, "trainId": 98},
|
| 126 |
+
{
|
| 127 |
+
"name": "aircraft carrier, carrier, flattop, attack aircraft carrier",
|
| 128 |
+
"id": 3144,
|
| 129 |
+
"trainId": 99,
|
| 130 |
+
},
|
| 131 |
+
{"name": "buffet, counter, sideboard", "id": 308, "trainId": 100},
|
| 132 |
+
{"name": "hovel, hut, hutch, shack, shanty", "id": 1282, "trainId": 101},
|
| 133 |
+
{"name": "apparel, wearing apparel, dress, clothes", "id": 38, "trainId": 102},
|
| 134 |
+
{"name": "minibike, motorbike", "id": 1563, "trainId": 103},
|
| 135 |
+
{"name": "animal, animate being, beast, brute, creature, fauna", "id": 29, "trainId": 104},
|
| 136 |
+
{"name": "chandelier, pendant, pendent", "id": 480, "trainId": 105},
|
| 137 |
+
{"name": "step, stair", "id": 2569, "trainId": 106},
|
| 138 |
+
{"name": "booth, cubicle, stall, kiosk", "id": 247, "trainId": 107},
|
| 139 |
+
{"name": "bicycle, bike, wheel, cycle", "id": 187, "trainId": 108},
|
| 140 |
+
{"name": "doorframe, doorcase", "id": 778, "trainId": 109},
|
| 141 |
+
{"name": "sconce", "id": 2243, "trainId": 110},
|
| 142 |
+
{"name": "pond", "id": 1941, "trainId": 111},
|
| 143 |
+
{"name": "trade name, brand name, brand, marque", "id": 2833, "trainId": 112},
|
| 144 |
+
{"name": "bannister, banister, balustrade, balusters, handrail", "id": 120, "trainId": 113},
|
| 145 |
+
{"name": "bag", "id": 95, "trainId": 114},
|
| 146 |
+
{"name": "traffic light, traffic signal, stoplight", "id": 2836, "trainId": 115},
|
| 147 |
+
{"name": "gazebo", "id": 1087, "trainId": 116},
|
| 148 |
+
{"name": "escalator, moving staircase, moving stairway", "id": 868, "trainId": 117},
|
| 149 |
+
{"name": "land, ground, soil", "id": 1401, "trainId": 118},
|
| 150 |
+
{"name": "board, plank", "id": 220, "trainId": 119},
|
| 151 |
+
{"name": "arcade machine", "id": 47, "trainId": 120},
|
| 152 |
+
{"name": "eiderdown, duvet, continental quilt", "id": 843, "trainId": 121},
|
| 153 |
+
{"name": "bar", "id": 123, "trainId": 122},
|
| 154 |
+
{"name": "stall, stand, sales booth", "id": 2537, "trainId": 123},
|
| 155 |
+
{"name": "playground", "id": 1927, "trainId": 124},
|
| 156 |
+
{"name": "ship", "id": 2337, "trainId": 125},
|
| 157 |
+
{"name": "ottoman, pouf, pouffe, puff, hassock", "id": 1702, "trainId": 126},
|
| 158 |
+
{
|
| 159 |
+
"name": "ashcan, trash can, garbage can, wastebin, ash bin, ash-bin, ashbin, dustbin, trash barrel, trash bin",
|
| 160 |
+
"id": 64,
|
| 161 |
+
"trainId": 127,
|
| 162 |
+
},
|
| 163 |
+
{"name": "bottle", "id": 249, "trainId": 128},
|
| 164 |
+
{"name": "cradle", "id": 642, "trainId": 129},
|
| 165 |
+
{"name": "pot, flowerpot", "id": 1981, "trainId": 130},
|
| 166 |
+
{
|
| 167 |
+
"name": "conveyer belt, conveyor belt, conveyer, conveyor, transporter",
|
| 168 |
+
"id": 609,
|
| 169 |
+
"trainId": 131,
|
| 170 |
+
},
|
| 171 |
+
{"name": "train, railroad train", "id": 2840, "trainId": 132},
|
| 172 |
+
{"name": "stool", "id": 2586, "trainId": 133},
|
| 173 |
+
{"name": "lake", "id": 1393, "trainId": 134},
|
| 174 |
+
{"name": "tank, storage tank", "id": 2704, "trainId": 135},
|
| 175 |
+
{"name": "ice, water ice", "id": 1304, "trainId": 136},
|
| 176 |
+
{"name": "basket, handbasket", "id": 146, "trainId": 137},
|
| 177 |
+
{"name": "manhole", "id": 1494, "trainId": 138},
|
| 178 |
+
{"name": "tent, collapsible shelter", "id": 2739, "trainId": 139},
|
| 179 |
+
{"name": "canopy", "id": 389, "trainId": 140},
|
| 180 |
+
{"name": "microwave, microwave oven", "id": 1551, "trainId": 141},
|
| 181 |
+
{"name": "barrel, cask", "id": 131, "trainId": 142},
|
| 182 |
+
{"name": "dirt track", "id": 738, "trainId": 143},
|
| 183 |
+
{"name": "beam", "id": 161, "trainId": 144},
|
| 184 |
+
{"name": "dishwasher, dish washer, dishwashing machine", "id": 747, "trainId": 145},
|
| 185 |
+
{"name": "plate", "id": 1919, "trainId": 146},
|
| 186 |
+
{"name": "screen, crt screen", "id": 3109, "trainId": 147},
|
| 187 |
+
{"name": "ruins", "id": 2179, "trainId": 148},
|
| 188 |
+
{"name": "washer, automatic washer, washing machine", "id": 2989, "trainId": 149},
|
| 189 |
+
{"name": "blanket, cover", "id": 206, "trainId": 150},
|
| 190 |
+
{"name": "plaything, toy", "id": 1930, "trainId": 151},
|
| 191 |
+
{"name": "food, solid food", "id": 1002, "trainId": 152},
|
| 192 |
+
{"name": "screen, silver screen, projection screen", "id": 2254, "trainId": 153},
|
| 193 |
+
{"name": "oven", "id": 1708, "trainId": 154},
|
| 194 |
+
{"name": "stage", "id": 2526, "trainId": 155},
|
| 195 |
+
{"name": "beacon, lighthouse, beacon light, pharos", "id": 160, "trainId": 156},
|
| 196 |
+
{"name": "umbrella", "id": 2901, "trainId": 157},
|
| 197 |
+
{"name": "sculpture", "id": 2262, "trainId": 158},
|
| 198 |
+
{"name": "aqueduct", "id": 44, "trainId": 159},
|
| 199 |
+
{"name": "container", "id": 597, "trainId": 160},
|
| 200 |
+
{"name": "scaffolding, staging", "id": 2235, "trainId": 161},
|
| 201 |
+
{"name": "hood, exhaust hood", "id": 1260, "trainId": 162},
|
| 202 |
+
{"name": "curb, curbing, kerb", "id": 682, "trainId": 163},
|
| 203 |
+
{"name": "roller coaster", "id": 2151, "trainId": 164},
|
| 204 |
+
{"name": "horse, equus caballus", "id": 3107, "trainId": 165},
|
| 205 |
+
{"name": "catwalk", "id": 432, "trainId": 166},
|
| 206 |
+
{"name": "glass, drinking glass", "id": 1098, "trainId": 167},
|
| 207 |
+
{"name": "vase", "id": 2932, "trainId": 168},
|
| 208 |
+
{"name": "central reservation", "id": 461, "trainId": 169},
|
| 209 |
+
{"name": "carousel", "id": 410, "trainId": 170},
|
| 210 |
+
{"name": "radiator", "id": 2046, "trainId": 171},
|
| 211 |
+
{"name": "closet", "id": 533, "trainId": 172},
|
| 212 |
+
{"name": "machine", "id": 1481, "trainId": 173},
|
| 213 |
+
{"name": "pier, wharf, wharfage, dock", "id": 1858, "trainId": 174},
|
| 214 |
+
{"name": "fan", "id": 894, "trainId": 175},
|
| 215 |
+
{"name": "inflatable bounce game", "id": 1322, "trainId": 176},
|
| 216 |
+
{"name": "pitch", "id": 1891, "trainId": 177},
|
| 217 |
+
{"name": "paper", "id": 1756, "trainId": 178},
|
| 218 |
+
{"name": "arcade, colonnade", "id": 49, "trainId": 179},
|
| 219 |
+
{"name": "hot tub", "id": 1272, "trainId": 180},
|
| 220 |
+
{"name": "helicopter", "id": 1229, "trainId": 181},
|
| 221 |
+
{"name": "tray", "id": 2850, "trainId": 182},
|
| 222 |
+
{"name": "partition, divider", "id": 1784, "trainId": 183},
|
| 223 |
+
{"name": "vineyard", "id": 2962, "trainId": 184},
|
| 224 |
+
{"name": "bowl", "id": 259, "trainId": 185},
|
| 225 |
+
{"name": "bullring", "id": 319, "trainId": 186},
|
| 226 |
+
{"name": "flag", "id": 954, "trainId": 187},
|
| 227 |
+
{"name": "pot", "id": 1974, "trainId": 188},
|
| 228 |
+
{"name": "footbridge, overcrossing, pedestrian bridge", "id": 1013, "trainId": 189},
|
| 229 |
+
{"name": "shower", "id": 2356, "trainId": 190},
|
| 230 |
+
{"name": "bag, traveling bag, travelling bag, grip, suitcase", "id": 97, "trainId": 191},
|
| 231 |
+
{"name": "bulletin board, notice board", "id": 318, "trainId": 192},
|
| 232 |
+
{"name": "confessional booth", "id": 592, "trainId": 193},
|
| 233 |
+
{"name": "trunk, tree trunk, bole", "id": 2885, "trainId": 194},
|
| 234 |
+
{"name": "forest", "id": 1017, "trainId": 195},
|
| 235 |
+
{"name": "elevator door", "id": 851, "trainId": 196},
|
| 236 |
+
{"name": "laptop, laptop computer", "id": 1407, "trainId": 197},
|
| 237 |
+
{"name": "instrument panel", "id": 1332, "trainId": 198},
|
| 238 |
+
{"name": "bucket, pail", "id": 303, "trainId": 199},
|
| 239 |
+
{"name": "tapestry, tapis", "id": 2714, "trainId": 200},
|
| 240 |
+
{"name": "platform", "id": 1924, "trainId": 201},
|
| 241 |
+
{"name": "jacket", "id": 1346, "trainId": 202},
|
| 242 |
+
{"name": "gate", "id": 1081, "trainId": 203},
|
| 243 |
+
{"name": "monitor, monitoring device", "id": 1583, "trainId": 204},
|
| 244 |
+
{
|
| 245 |
+
"name": "telephone booth, phone booth, call box, telephone box, telephone kiosk",
|
| 246 |
+
"id": 2727,
|
| 247 |
+
"trainId": 205,
|
| 248 |
+
},
|
| 249 |
+
{"name": "spotlight, spot", "id": 2509, "trainId": 206},
|
| 250 |
+
{"name": "ring", "id": 2123, "trainId": 207},
|
| 251 |
+
{"name": "control panel", "id": 602, "trainId": 208},
|
| 252 |
+
{"name": "blackboard, chalkboard", "id": 202, "trainId": 209},
|
| 253 |
+
{"name": "air conditioner, air conditioning", "id": 10, "trainId": 210},
|
| 254 |
+
{"name": "chest", "id": 490, "trainId": 211},
|
| 255 |
+
{"name": "clock", "id": 530, "trainId": 212},
|
| 256 |
+
{"name": "sand dune", "id": 2213, "trainId": 213},
|
| 257 |
+
{"name": "pipe, pipage, piping", "id": 1884, "trainId": 214},
|
| 258 |
+
{"name": "vault", "id": 2934, "trainId": 215},
|
| 259 |
+
{"name": "table football", "id": 2687, "trainId": 216},
|
| 260 |
+
{"name": "cannon", "id": 387, "trainId": 217},
|
| 261 |
+
{"name": "swimming pool, swimming bath, natatorium", "id": 2668, "trainId": 218},
|
| 262 |
+
{"name": "fluorescent, fluorescent fixture", "id": 982, "trainId": 219},
|
| 263 |
+
{"name": "statue", "id": 2547, "trainId": 220},
|
| 264 |
+
{
|
| 265 |
+
"name": "loudspeaker, speaker, speaker unit, loudspeaker system, speaker system",
|
| 266 |
+
"id": 1474,
|
| 267 |
+
"trainId": 221,
|
| 268 |
+
},
|
| 269 |
+
{"name": "exhibitor", "id": 877, "trainId": 222},
|
| 270 |
+
{"name": "ladder", "id": 1391, "trainId": 223},
|
| 271 |
+
{"name": "carport", "id": 414, "trainId": 224},
|
| 272 |
+
{"name": "dam", "id": 698, "trainId": 225},
|
| 273 |
+
{"name": "pulpit", "id": 2019, "trainId": 226},
|
| 274 |
+
{"name": "skylight, fanlight", "id": 2422, "trainId": 227},
|
| 275 |
+
{"name": "water tower", "id": 3010, "trainId": 228},
|
| 276 |
+
{"name": "grill, grille, grillwork", "id": 1139, "trainId": 229},
|
| 277 |
+
{"name": "display board", "id": 753, "trainId": 230},
|
| 278 |
+
{"name": "pane, pane of glass, window glass", "id": 1747, "trainId": 231},
|
| 279 |
+
{"name": "rubbish, trash, scrap", "id": 2175, "trainId": 232},
|
| 280 |
+
{"name": "ice rink", "id": 1301, "trainId": 233},
|
| 281 |
+
{"name": "fruit", "id": 1033, "trainId": 234},
|
| 282 |
+
{"name": "patio", "id": 1789, "trainId": 235},
|
| 283 |
+
{"name": "vending machine", "id": 2939, "trainId": 236},
|
| 284 |
+
{"name": "telephone, phone, telephone set", "id": 2730, "trainId": 237},
|
| 285 |
+
{"name": "net", "id": 1652, "trainId": 238},
|
| 286 |
+
{
|
| 287 |
+
"name": "backpack, back pack, knapsack, packsack, rucksack, haversack",
|
| 288 |
+
"id": 90,
|
| 289 |
+
"trainId": 239,
|
| 290 |
+
},
|
| 291 |
+
{"name": "jar", "id": 1349, "trainId": 240},
|
| 292 |
+
{"name": "track", "id": 2830, "trainId": 241},
|
| 293 |
+
{"name": "magazine", "id": 1485, "trainId": 242},
|
| 294 |
+
{"name": "shutter", "id": 2370, "trainId": 243},
|
| 295 |
+
{"name": "roof", "id": 2155, "trainId": 244},
|
| 296 |
+
{"name": "banner, streamer", "id": 118, "trainId": 245},
|
| 297 |
+
{"name": "landfill", "id": 1402, "trainId": 246},
|
| 298 |
+
{"name": "post", "id": 1957, "trainId": 247},
|
| 299 |
+
{"name": "altarpiece, reredos", "id": 3130, "trainId": 248},
|
| 300 |
+
{"name": "hat, chapeau, lid", "id": 1197, "trainId": 249},
|
| 301 |
+
{"name": "arch, archway", "id": 52, "trainId": 250},
|
| 302 |
+
{"name": "table game", "id": 2688, "trainId": 251},
|
| 303 |
+
{"name": "bag, handbag, pocketbook, purse", "id": 96, "trainId": 252},
|
| 304 |
+
{"name": "document, written document, papers", "id": 762, "trainId": 253},
|
| 305 |
+
{"name": "dome", "id": 772, "trainId": 254},
|
| 306 |
+
{"name": "pier", "id": 1857, "trainId": 255},
|
| 307 |
+
{"name": "shanties", "id": 2315, "trainId": 256},
|
| 308 |
+
{"name": "forecourt", "id": 1016, "trainId": 257},
|
| 309 |
+
{"name": "crane", "id": 643, "trainId": 258},
|
| 310 |
+
{"name": "dog, domestic dog, canis familiaris", "id": 3105, "trainId": 259},
|
| 311 |
+
{"name": "piano, pianoforte, forte-piano", "id": 1849, "trainId": 260},
|
| 312 |
+
{"name": "drawing", "id": 791, "trainId": 261},
|
| 313 |
+
{"name": "cabin", "id": 349, "trainId": 262},
|
| 314 |
+
{
|
| 315 |
+
"name": "ad, advertisement, advertizement, advertising, advertizing, advert",
|
| 316 |
+
"id": 6,
|
| 317 |
+
"trainId": 263,
|
| 318 |
+
},
|
| 319 |
+
{"name": "amphitheater, amphitheatre, coliseum", "id": 3114, "trainId": 264},
|
| 320 |
+
{"name": "monument", "id": 1587, "trainId": 265},
|
| 321 |
+
{"name": "henhouse", "id": 1233, "trainId": 266},
|
| 322 |
+
{"name": "cockpit", "id": 559, "trainId": 267},
|
| 323 |
+
{"name": "heater, warmer", "id": 1223, "trainId": 268},
|
| 324 |
+
{"name": "windmill, aerogenerator, wind generator", "id": 3049, "trainId": 269},
|
| 325 |
+
{"name": "pool", "id": 1943, "trainId": 270},
|
| 326 |
+
{"name": "elevator, lift", "id": 853, "trainId": 271},
|
| 327 |
+
{"name": "decoration, ornament, ornamentation", "id": 709, "trainId": 272},
|
| 328 |
+
{"name": "labyrinth", "id": 1390, "trainId": 273},
|
| 329 |
+
{"name": "text, textual matter", "id": 2748, "trainId": 274},
|
| 330 |
+
{"name": "printer", "id": 2007, "trainId": 275},
|
| 331 |
+
{"name": "mezzanine, first balcony", "id": 1546, "trainId": 276},
|
| 332 |
+
{"name": "mattress", "id": 1513, "trainId": 277},
|
| 333 |
+
{"name": "straw", "id": 2600, "trainId": 278},
|
| 334 |
+
{"name": "stalls", "id": 2538, "trainId": 279},
|
| 335 |
+
{"name": "patio, terrace", "id": 1790, "trainId": 280},
|
| 336 |
+
{"name": "billboard, hoarding", "id": 194, "trainId": 281},
|
| 337 |
+
{"name": "bus stop", "id": 326, "trainId": 282},
|
| 338 |
+
{"name": "trouser, pant", "id": 2877, "trainId": 283},
|
| 339 |
+
{"name": "console table, console", "id": 594, "trainId": 284},
|
| 340 |
+
{"name": "rack", "id": 2036, "trainId": 285},
|
| 341 |
+
{"name": "notebook", "id": 1662, "trainId": 286},
|
| 342 |
+
{"name": "shrine", "id": 2366, "trainId": 287},
|
| 343 |
+
{"name": "pantry", "id": 1754, "trainId": 288},
|
| 344 |
+
{"name": "cart", "id": 418, "trainId": 289},
|
| 345 |
+
{"name": "steam shovel", "id": 2553, "trainId": 290},
|
| 346 |
+
{"name": "porch", "id": 1951, "trainId": 291},
|
| 347 |
+
{"name": "postbox, mailbox, letter box", "id": 1963, "trainId": 292},
|
| 348 |
+
{"name": "figurine, statuette", "id": 918, "trainId": 293},
|
| 349 |
+
{"name": "recycling bin", "id": 2086, "trainId": 294},
|
| 350 |
+
{"name": "folding screen", "id": 997, "trainId": 295},
|
| 351 |
+
{"name": "telescope", "id": 2731, "trainId": 296},
|
| 352 |
+
{"name": "deck chair, beach chair", "id": 704, "trainId": 297},
|
| 353 |
+
{"name": "kennel", "id": 1365, "trainId": 298},
|
| 354 |
+
{"name": "coffee maker", "id": 569, "trainId": 299},
|
| 355 |
+
{"name": "altar, communion table, lord's table", "id": 3108, "trainId": 300},
|
| 356 |
+
{"name": "fish", "id": 948, "trainId": 301},
|
| 357 |
+
{"name": "easel", "id": 839, "trainId": 302},
|
| 358 |
+
{"name": "artificial golf green", "id": 63, "trainId": 303},
|
| 359 |
+
{"name": "iceberg", "id": 1305, "trainId": 304},
|
| 360 |
+
{"name": "candlestick, candle holder", "id": 378, "trainId": 305},
|
| 361 |
+
{"name": "shower stall, shower bath", "id": 2362, "trainId": 306},
|
| 362 |
+
{"name": "television stand", "id": 2734, "trainId": 307},
|
| 363 |
+
{
|
| 364 |
+
"name": "wall socket, wall plug, electric outlet, electrical outlet, outlet, electric receptacle",
|
| 365 |
+
"id": 2982,
|
| 366 |
+
"trainId": 308,
|
| 367 |
+
},
|
| 368 |
+
{"name": "skeleton", "id": 2398, "trainId": 309},
|
| 369 |
+
{"name": "grand piano, grand", "id": 1119, "trainId": 310},
|
| 370 |
+
{"name": "candy, confect", "id": 382, "trainId": 311},
|
| 371 |
+
{"name": "grille door", "id": 1141, "trainId": 312},
|
| 372 |
+
{"name": "pedestal, plinth, footstall", "id": 1805, "trainId": 313},
|
| 373 |
+
{"name": "jersey, t-shirt, tee shirt", "id": 3102, "trainId": 314},
|
| 374 |
+
{"name": "shoe", "id": 2341, "trainId": 315},
|
| 375 |
+
{"name": "gravestone, headstone, tombstone", "id": 1131, "trainId": 316},
|
| 376 |
+
{"name": "shanty", "id": 2316, "trainId": 317},
|
| 377 |
+
{"name": "structure", "id": 2626, "trainId": 318},
|
| 378 |
+
{"name": "rocking chair, rocker", "id": 3104, "trainId": 319},
|
| 379 |
+
{"name": "bird", "id": 198, "trainId": 320},
|
| 380 |
+
{"name": "place mat", "id": 1896, "trainId": 321},
|
| 381 |
+
{"name": "tomb", "id": 2800, "trainId": 322},
|
| 382 |
+
{"name": "big top", "id": 190, "trainId": 323},
|
| 383 |
+
{"name": "gas pump, gasoline pump, petrol pump, island dispenser", "id": 3131, "trainId": 324},
|
| 384 |
+
{"name": "lockers", "id": 1463, "trainId": 325},
|
| 385 |
+
{"name": "cage", "id": 357, "trainId": 326},
|
| 386 |
+
{"name": "finger", "id": 929, "trainId": 327},
|
| 387 |
+
{"name": "bleachers", "id": 209, "trainId": 328},
|
| 388 |
+
{"name": "ferris wheel", "id": 912, "trainId": 329},
|
| 389 |
+
{"name": "hairdresser chair", "id": 1164, "trainId": 330},
|
| 390 |
+
{"name": "mat", "id": 1509, "trainId": 331},
|
| 391 |
+
{"name": "stands", "id": 2539, "trainId": 332},
|
| 392 |
+
{"name": "aquarium, fish tank, marine museum", "id": 3116, "trainId": 333},
|
| 393 |
+
{"name": "streetcar, tram, tramcar, trolley, trolley car", "id": 2615, "trainId": 334},
|
| 394 |
+
{"name": "napkin, table napkin, serviette", "id": 1644, "trainId": 335},
|
| 395 |
+
{"name": "dummy", "id": 818, "trainId": 336},
|
| 396 |
+
{"name": "booklet, brochure, folder, leaflet, pamphlet", "id": 242, "trainId": 337},
|
| 397 |
+
{"name": "sand trap", "id": 2217, "trainId": 338},
|
| 398 |
+
{"name": "shop, store", "id": 2347, "trainId": 339},
|
| 399 |
+
{"name": "table cloth", "id": 2686, "trainId": 340},
|
| 400 |
+
{"name": "service station", "id": 2300, "trainId": 341},
|
| 401 |
+
{"name": "coffin", "id": 572, "trainId": 342},
|
| 402 |
+
{"name": "drawer", "id": 789, "trainId": 343},
|
| 403 |
+
{"name": "cages", "id": 358, "trainId": 344},
|
| 404 |
+
{"name": "slot machine, coin machine", "id": 2443, "trainId": 345},
|
| 405 |
+
{"name": "balcony", "id": 101, "trainId": 346},
|
| 406 |
+
{"name": "volleyball court", "id": 2969, "trainId": 347},
|
| 407 |
+
{"name": "table tennis", "id": 2692, "trainId": 348},
|
| 408 |
+
{"name": "control table", "id": 606, "trainId": 349},
|
| 409 |
+
{"name": "shirt", "id": 2339, "trainId": 350},
|
| 410 |
+
{"name": "merchandise, ware, product", "id": 1533, "trainId": 351},
|
| 411 |
+
{"name": "railway", "id": 2060, "trainId": 352},
|
| 412 |
+
{"name": "parterre", "id": 1782, "trainId": 353},
|
| 413 |
+
{"name": "chimney", "id": 495, "trainId": 354},
|
| 414 |
+
{"name": "can, tin, tin can", "id": 371, "trainId": 355},
|
| 415 |
+
{"name": "tanks", "id": 2707, "trainId": 356},
|
| 416 |
+
{"name": "fabric, cloth, material, textile", "id": 889, "trainId": 357},
|
| 417 |
+
{"name": "alga, algae", "id": 3156, "trainId": 358},
|
| 418 |
+
{"name": "system", "id": 2683, "trainId": 359},
|
| 419 |
+
{"name": "map", "id": 1499, "trainId": 360},
|
| 420 |
+
{"name": "greenhouse", "id": 1135, "trainId": 361},
|
| 421 |
+
{"name": "mug", "id": 1619, "trainId": 362},
|
| 422 |
+
{"name": "barbecue", "id": 125, "trainId": 363},
|
| 423 |
+
{"name": "trailer", "id": 2838, "trainId": 364},
|
| 424 |
+
{"name": "toilet tissue, toilet paper, bathroom tissue", "id": 2792, "trainId": 365},
|
| 425 |
+
{"name": "organ", "id": 1695, "trainId": 366},
|
| 426 |
+
{"name": "dishrag, dishcloth", "id": 746, "trainId": 367},
|
| 427 |
+
{"name": "island", "id": 1343, "trainId": 368},
|
| 428 |
+
{"name": "keyboard", "id": 1370, "trainId": 369},
|
| 429 |
+
{"name": "trench", "id": 2858, "trainId": 370},
|
| 430 |
+
{"name": "basket, basketball hoop, hoop", "id": 145, "trainId": 371},
|
| 431 |
+
{"name": "steering wheel, wheel", "id": 2565, "trainId": 372},
|
| 432 |
+
{"name": "pitcher, ewer", "id": 1892, "trainId": 373},
|
| 433 |
+
{"name": "goal", "id": 1103, "trainId": 374},
|
| 434 |
+
{"name": "bread, breadstuff, staff of life", "id": 286, "trainId": 375},
|
| 435 |
+
{"name": "beds", "id": 170, "trainId": 376},
|
| 436 |
+
{"name": "wood", "id": 3073, "trainId": 377},
|
| 437 |
+
{"name": "file cabinet", "id": 922, "trainId": 378},
|
| 438 |
+
{"name": "newspaper, paper", "id": 1655, "trainId": 379},
|
| 439 |
+
{"name": "motorboat", "id": 1602, "trainId": 380},
|
| 440 |
+
{"name": "rope", "id": 2160, "trainId": 381},
|
| 441 |
+
{"name": "guitar", "id": 1151, "trainId": 382},
|
| 442 |
+
{"name": "rubble", "id": 2176, "trainId": 383},
|
| 443 |
+
{"name": "scarf", "id": 2239, "trainId": 384},
|
| 444 |
+
{"name": "barrels", "id": 132, "trainId": 385},
|
| 445 |
+
{"name": "cap", "id": 394, "trainId": 386},
|
| 446 |
+
{"name": "leaves", "id": 1424, "trainId": 387},
|
| 447 |
+
{"name": "control tower", "id": 607, "trainId": 388},
|
| 448 |
+
{"name": "dashboard", "id": 700, "trainId": 389},
|
| 449 |
+
{"name": "bandstand", "id": 116, "trainId": 390},
|
| 450 |
+
{"name": "lectern", "id": 1425, "trainId": 391},
|
| 451 |
+
{"name": "switch, electric switch, electrical switch", "id": 2676, "trainId": 392},
|
| 452 |
+
{"name": "baseboard, mopboard, skirting board", "id": 141, "trainId": 393},
|
| 453 |
+
{"name": "shower room", "id": 2360, "trainId": 394},
|
| 454 |
+
{"name": "smoke", "id": 2449, "trainId": 395},
|
| 455 |
+
{"name": "faucet, spigot", "id": 897, "trainId": 396},
|
| 456 |
+
{"name": "bulldozer", "id": 317, "trainId": 397},
|
| 457 |
+
{"name": "saucepan", "id": 2228, "trainId": 398},
|
| 458 |
+
{"name": "shops", "id": 2351, "trainId": 399},
|
| 459 |
+
{"name": "meter", "id": 1543, "trainId": 400},
|
| 460 |
+
{"name": "crevasse", "id": 656, "trainId": 401},
|
| 461 |
+
{"name": "gear", "id": 1088, "trainId": 402},
|
| 462 |
+
{"name": "candelabrum, candelabra", "id": 373, "trainId": 403},
|
| 463 |
+
{"name": "sofa bed", "id": 2472, "trainId": 404},
|
| 464 |
+
{"name": "tunnel", "id": 2892, "trainId": 405},
|
| 465 |
+
{"name": "pallet", "id": 1740, "trainId": 406},
|
| 466 |
+
{"name": "wire, conducting wire", "id": 3067, "trainId": 407},
|
| 467 |
+
{"name": "kettle, boiler", "id": 1367, "trainId": 408},
|
| 468 |
+
{"name": "bidet", "id": 188, "trainId": 409},
|
| 469 |
+
{
|
| 470 |
+
"name": "baby buggy, baby carriage, carriage, perambulator, pram, stroller, go-cart, pushchair, pusher",
|
| 471 |
+
"id": 79,
|
| 472 |
+
"trainId": 410,
|
| 473 |
+
},
|
| 474 |
+
{"name": "music stand", "id": 1633, "trainId": 411},
|
| 475 |
+
{"name": "pipe, tube", "id": 1885, "trainId": 412},
|
| 476 |
+
{"name": "cup", "id": 677, "trainId": 413},
|
| 477 |
+
{"name": "parking meter", "id": 1779, "trainId": 414},
|
| 478 |
+
{"name": "ice hockey rink", "id": 1297, "trainId": 415},
|
| 479 |
+
{"name": "shelter", "id": 2334, "trainId": 416},
|
| 480 |
+
{"name": "weeds", "id": 3027, "trainId": 417},
|
| 481 |
+
{"name": "temple", "id": 2735, "trainId": 418},
|
| 482 |
+
{"name": "patty, cake", "id": 1791, "trainId": 419},
|
| 483 |
+
{"name": "ski slope", "id": 2405, "trainId": 420},
|
| 484 |
+
{"name": "panel", "id": 1748, "trainId": 421},
|
| 485 |
+
{"name": "wallet", "id": 2983, "trainId": 422},
|
| 486 |
+
{"name": "wheel", "id": 3035, "trainId": 423},
|
| 487 |
+
{"name": "towel rack, towel horse", "id": 2824, "trainId": 424},
|
| 488 |
+
{"name": "roundabout", "id": 2168, "trainId": 425},
|
| 489 |
+
{"name": "canister, cannister, tin", "id": 385, "trainId": 426},
|
| 490 |
+
{"name": "rod", "id": 2148, "trainId": 427},
|
| 491 |
+
{"name": "soap dispenser", "id": 2465, "trainId": 428},
|
| 492 |
+
{"name": "bell", "id": 175, "trainId": 429},
|
| 493 |
+
{"name": "canvas", "id": 390, "trainId": 430},
|
| 494 |
+
{"name": "box office, ticket office, ticket booth", "id": 268, "trainId": 431},
|
| 495 |
+
{"name": "teacup", "id": 2722, "trainId": 432},
|
| 496 |
+
{"name": "trellis", "id": 2857, "trainId": 433},
|
| 497 |
+
{"name": "workbench", "id": 3088, "trainId": 434},
|
| 498 |
+
{"name": "valley, vale", "id": 2926, "trainId": 435},
|
| 499 |
+
{"name": "toaster", "id": 2782, "trainId": 436},
|
| 500 |
+
{"name": "knife", "id": 1378, "trainId": 437},
|
| 501 |
+
{"name": "podium", "id": 1934, "trainId": 438},
|
| 502 |
+
{"name": "ramp", "id": 2072, "trainId": 439},
|
| 503 |
+
{"name": "tumble dryer", "id": 2889, "trainId": 440},
|
| 504 |
+
{"name": "fireplug, fire hydrant, plug", "id": 944, "trainId": 441},
|
| 505 |
+
{"name": "gym shoe, sneaker, tennis shoe", "id": 1158, "trainId": 442},
|
| 506 |
+
{"name": "lab bench", "id": 1383, "trainId": 443},
|
| 507 |
+
{"name": "equipment", "id": 867, "trainId": 444},
|
| 508 |
+
{"name": "rocky formation", "id": 2145, "trainId": 445},
|
| 509 |
+
{"name": "plastic", "id": 1915, "trainId": 446},
|
| 510 |
+
{"name": "calendar", "id": 361, "trainId": 447},
|
| 511 |
+
{"name": "caravan", "id": 402, "trainId": 448},
|
| 512 |
+
{"name": "check-in-desk", "id": 482, "trainId": 449},
|
| 513 |
+
{"name": "ticket counter", "id": 2761, "trainId": 450},
|
| 514 |
+
{"name": "brush", "id": 300, "trainId": 451},
|
| 515 |
+
{"name": "mill", "id": 1554, "trainId": 452},
|
| 516 |
+
{"name": "covered bridge", "id": 636, "trainId": 453},
|
| 517 |
+
{"name": "bowling alley", "id": 260, "trainId": 454},
|
| 518 |
+
{"name": "hanger", "id": 1186, "trainId": 455},
|
| 519 |
+
{"name": "excavator", "id": 871, "trainId": 456},
|
| 520 |
+
{"name": "trestle", "id": 2859, "trainId": 457},
|
| 521 |
+
{"name": "revolving door", "id": 2103, "trainId": 458},
|
| 522 |
+
{"name": "blast furnace", "id": 208, "trainId": 459},
|
| 523 |
+
{"name": "scale, weighing machine", "id": 2236, "trainId": 460},
|
| 524 |
+
{"name": "projector", "id": 2012, "trainId": 461},
|
| 525 |
+
{"name": "soap", "id": 2462, "trainId": 462},
|
| 526 |
+
{"name": "locker", "id": 1462, "trainId": 463},
|
| 527 |
+
{"name": "tractor", "id": 2832, "trainId": 464},
|
| 528 |
+
{"name": "stretcher", "id": 2617, "trainId": 465},
|
| 529 |
+
{"name": "frame", "id": 1024, "trainId": 466},
|
| 530 |
+
{"name": "grating", "id": 1129, "trainId": 467},
|
| 531 |
+
{"name": "alembic", "id": 18, "trainId": 468},
|
| 532 |
+
{"name": "candle, taper, wax light", "id": 376, "trainId": 469},
|
| 533 |
+
{"name": "barrier", "id": 134, "trainId": 470},
|
| 534 |
+
{"name": "cardboard", "id": 407, "trainId": 471},
|
| 535 |
+
{"name": "cave", "id": 434, "trainId": 472},
|
| 536 |
+
{"name": "puddle", "id": 2017, "trainId": 473},
|
| 537 |
+
{"name": "tarp", "id": 2717, "trainId": 474},
|
| 538 |
+
{"name": "price tag", "id": 2005, "trainId": 475},
|
| 539 |
+
{"name": "watchtower", "id": 2993, "trainId": 476},
|
| 540 |
+
{"name": "meters", "id": 1545, "trainId": 477},
|
| 541 |
+
{
|
| 542 |
+
"name": "light bulb, lightbulb, bulb, incandescent lamp, electric light, electric-light bulb",
|
| 543 |
+
"id": 1445,
|
| 544 |
+
"trainId": 478,
|
| 545 |
+
},
|
| 546 |
+
{"name": "tracks", "id": 2831, "trainId": 479},
|
| 547 |
+
{"name": "hair dryer", "id": 1161, "trainId": 480},
|
| 548 |
+
{"name": "skirt", "id": 2411, "trainId": 481},
|
| 549 |
+
{"name": "viaduct", "id": 2949, "trainId": 482},
|
| 550 |
+
{"name": "paper towel", "id": 1769, "trainId": 483},
|
| 551 |
+
{"name": "coat", "id": 552, "trainId": 484},
|
| 552 |
+
{"name": "sheet", "id": 2327, "trainId": 485},
|
| 553 |
+
{"name": "fire extinguisher, extinguisher, asphyxiator", "id": 939, "trainId": 486},
|
| 554 |
+
{"name": "water wheel", "id": 3013, "trainId": 487},
|
| 555 |
+
{"name": "pottery, clayware", "id": 1986, "trainId": 488},
|
| 556 |
+
{"name": "magazine rack", "id": 1486, "trainId": 489},
|
| 557 |
+
{"name": "teapot", "id": 2723, "trainId": 490},
|
| 558 |
+
{"name": "microphone, mike", "id": 1549, "trainId": 491},
|
| 559 |
+
{"name": "support", "id": 2649, "trainId": 492},
|
| 560 |
+
{"name": "forklift", "id": 1020, "trainId": 493},
|
| 561 |
+
{"name": "canyon", "id": 392, "trainId": 494},
|
| 562 |
+
{"name": "cash register, register", "id": 422, "trainId": 495},
|
| 563 |
+
{"name": "leaf, leafage, foliage", "id": 1419, "trainId": 496},
|
| 564 |
+
{"name": "remote control, remote", "id": 2099, "trainId": 497},
|
| 565 |
+
{"name": "soap dish", "id": 2464, "trainId": 498},
|
| 566 |
+
{"name": "windshield, windscreen", "id": 3058, "trainId": 499},
|
| 567 |
+
{"name": "cat", "id": 430, "trainId": 500},
|
| 568 |
+
{"name": "cue, cue stick, pool cue, pool stick", "id": 675, "trainId": 501},
|
| 569 |
+
{"name": "vent, venthole, vent-hole, blowhole", "id": 2941, "trainId": 502},
|
| 570 |
+
{"name": "videos", "id": 2955, "trainId": 503},
|
| 571 |
+
{"name": "shovel", "id": 2355, "trainId": 504},
|
| 572 |
+
{"name": "eaves", "id": 840, "trainId": 505},
|
| 573 |
+
{"name": "antenna, aerial, transmitting aerial", "id": 32, "trainId": 506},
|
| 574 |
+
{"name": "shipyard", "id": 2338, "trainId": 507},
|
| 575 |
+
{"name": "hen, biddy", "id": 1232, "trainId": 508},
|
| 576 |
+
{"name": "traffic cone", "id": 2834, "trainId": 509},
|
| 577 |
+
{"name": "washing machines", "id": 2991, "trainId": 510},
|
| 578 |
+
{"name": "truck crane", "id": 2879, "trainId": 511},
|
| 579 |
+
{"name": "cds", "id": 444, "trainId": 512},
|
| 580 |
+
{"name": "niche", "id": 1657, "trainId": 513},
|
| 581 |
+
{"name": "scoreboard", "id": 2246, "trainId": 514},
|
| 582 |
+
{"name": "briefcase", "id": 296, "trainId": 515},
|
| 583 |
+
{"name": "boot", "id": 245, "trainId": 516},
|
| 584 |
+
{"name": "sweater, jumper", "id": 2661, "trainId": 517},
|
| 585 |
+
{"name": "hay", "id": 1202, "trainId": 518},
|
| 586 |
+
{"name": "pack", "id": 1714, "trainId": 519},
|
| 587 |
+
{"name": "bottle rack", "id": 251, "trainId": 520},
|
| 588 |
+
{"name": "glacier", "id": 1095, "trainId": 521},
|
| 589 |
+
{"name": "pergola", "id": 1828, "trainId": 522},
|
| 590 |
+
{"name": "building materials", "id": 311, "trainId": 523},
|
| 591 |
+
{"name": "television camera", "id": 2732, "trainId": 524},
|
| 592 |
+
{"name": "first floor", "id": 947, "trainId": 525},
|
| 593 |
+
{"name": "rifle", "id": 2115, "trainId": 526},
|
| 594 |
+
{"name": "tennis table", "id": 2738, "trainId": 527},
|
| 595 |
+
{"name": "stadium", "id": 2525, "trainId": 528},
|
| 596 |
+
{"name": "safety belt", "id": 2194, "trainId": 529},
|
| 597 |
+
{"name": "cover", "id": 634, "trainId": 530},
|
| 598 |
+
{"name": "dish rack", "id": 740, "trainId": 531},
|
| 599 |
+
{"name": "synthesizer", "id": 2682, "trainId": 532},
|
| 600 |
+
{"name": "pumpkin", "id": 2020, "trainId": 533},
|
| 601 |
+
{"name": "gutter", "id": 1156, "trainId": 534},
|
| 602 |
+
{"name": "fruit stand", "id": 1036, "trainId": 535},
|
| 603 |
+
{"name": "ice floe, floe", "id": 1295, "trainId": 536},
|
| 604 |
+
{"name": "handle, grip, handgrip, hold", "id": 1181, "trainId": 537},
|
| 605 |
+
{"name": "wheelchair", "id": 3037, "trainId": 538},
|
| 606 |
+
{"name": "mousepad, mouse mat", "id": 1614, "trainId": 539},
|
| 607 |
+
{"name": "diploma", "id": 736, "trainId": 540},
|
| 608 |
+
{"name": "fairground ride", "id": 893, "trainId": 541},
|
| 609 |
+
{"name": "radio", "id": 2047, "trainId": 542},
|
| 610 |
+
{"name": "hotplate", "id": 1274, "trainId": 543},
|
| 611 |
+
{"name": "junk", "id": 1361, "trainId": 544},
|
| 612 |
+
{"name": "wheelbarrow", "id": 3036, "trainId": 545},
|
| 613 |
+
{"name": "stream", "id": 2606, "trainId": 546},
|
| 614 |
+
{"name": "toll plaza", "id": 2797, "trainId": 547},
|
| 615 |
+
{"name": "punching bag", "id": 2022, "trainId": 548},
|
| 616 |
+
{"name": "trough", "id": 2876, "trainId": 549},
|
| 617 |
+
{"name": "throne", "id": 2758, "trainId": 550},
|
| 618 |
+
{"name": "chair desk", "id": 472, "trainId": 551},
|
| 619 |
+
{"name": "weighbridge", "id": 3028, "trainId": 552},
|
| 620 |
+
{"name": "extractor fan", "id": 882, "trainId": 553},
|
| 621 |
+
{"name": "hanging clothes", "id": 1189, "trainId": 554},
|
| 622 |
+
{"name": "dish, dish aerial, dish antenna, saucer", "id": 743, "trainId": 555},
|
| 623 |
+
{"name": "alarm clock, alarm", "id": 3122, "trainId": 556},
|
| 624 |
+
{"name": "ski lift", "id": 2401, "trainId": 557},
|
| 625 |
+
{"name": "chain", "id": 468, "trainId": 558},
|
| 626 |
+
{"name": "garage", "id": 1061, "trainId": 559},
|
| 627 |
+
{"name": "mechanical shovel", "id": 1523, "trainId": 560},
|
| 628 |
+
{"name": "wine rack", "id": 3059, "trainId": 561},
|
| 629 |
+
{"name": "tramway", "id": 2843, "trainId": 562},
|
| 630 |
+
{"name": "treadmill", "id": 2853, "trainId": 563},
|
| 631 |
+
{"name": "menu", "id": 1529, "trainId": 564},
|
| 632 |
+
{"name": "block", "id": 214, "trainId": 565},
|
| 633 |
+
{"name": "well", "id": 3032, "trainId": 566},
|
| 634 |
+
{"name": "witness stand", "id": 3071, "trainId": 567},
|
| 635 |
+
{"name": "branch", "id": 277, "trainId": 568},
|
| 636 |
+
{"name": "duck", "id": 813, "trainId": 569},
|
| 637 |
+
{"name": "casserole", "id": 426, "trainId": 570},
|
| 638 |
+
{"name": "frying pan", "id": 1039, "trainId": 571},
|
| 639 |
+
{"name": "desk organizer", "id": 727, "trainId": 572},
|
| 640 |
+
{"name": "mast", "id": 1508, "trainId": 573},
|
| 641 |
+
{"name": "spectacles, specs, eyeglasses, glasses", "id": 2490, "trainId": 574},
|
| 642 |
+
{"name": "service elevator", "id": 2299, "trainId": 575},
|
| 643 |
+
{"name": "dollhouse", "id": 768, "trainId": 576},
|
| 644 |
+
{"name": "hammock", "id": 1172, "trainId": 577},
|
| 645 |
+
{"name": "clothes hanging", "id": 537, "trainId": 578},
|
| 646 |
+
{"name": "photocopier", "id": 1847, "trainId": 579},
|
| 647 |
+
{"name": "notepad", "id": 1664, "trainId": 580},
|
| 648 |
+
{"name": "golf cart", "id": 1110, "trainId": 581},
|
| 649 |
+
{"name": "footpath", "id": 1014, "trainId": 582},
|
| 650 |
+
{"name": "cross", "id": 662, "trainId": 583},
|
| 651 |
+
{"name": "baptismal font", "id": 121, "trainId": 584},
|
| 652 |
+
{"name": "boiler", "id": 227, "trainId": 585},
|
| 653 |
+
{"name": "skip", "id": 2410, "trainId": 586},
|
| 654 |
+
{"name": "rotisserie", "id": 2165, "trainId": 587},
|
| 655 |
+
{"name": "tables", "id": 2696, "trainId": 588},
|
| 656 |
+
{"name": "water mill", "id": 3005, "trainId": 589},
|
| 657 |
+
{"name": "helmet", "id": 1231, "trainId": 590},
|
| 658 |
+
{"name": "cover curtain", "id": 635, "trainId": 591},
|
| 659 |
+
{"name": "brick", "id": 292, "trainId": 592},
|
| 660 |
+
{"name": "table runner", "id": 2690, "trainId": 593},
|
| 661 |
+
{"name": "ashtray", "id": 65, "trainId": 594},
|
| 662 |
+
{"name": "street box", "id": 2607, "trainId": 595},
|
| 663 |
+
{"name": "stick", "id": 2574, "trainId": 596},
|
| 664 |
+
{"name": "hangers", "id": 1188, "trainId": 597},
|
| 665 |
+
{"name": "cells", "id": 456, "trainId": 598},
|
| 666 |
+
{"name": "urinal", "id": 2913, "trainId": 599},
|
| 667 |
+
{"name": "centerpiece", "id": 459, "trainId": 600},
|
| 668 |
+
{"name": "portable fridge", "id": 1955, "trainId": 601},
|
| 669 |
+
{"name": "dvds", "id": 827, "trainId": 602},
|
| 670 |
+
{"name": "golf club", "id": 1111, "trainId": 603},
|
| 671 |
+
{"name": "skirting board", "id": 2412, "trainId": 604},
|
| 672 |
+
{"name": "water cooler", "id": 2997, "trainId": 605},
|
| 673 |
+
{"name": "clipboard", "id": 528, "trainId": 606},
|
| 674 |
+
{"name": "camera, photographic camera", "id": 366, "trainId": 607},
|
| 675 |
+
{"name": "pigeonhole", "id": 1863, "trainId": 608},
|
| 676 |
+
{"name": "chips", "id": 500, "trainId": 609},
|
| 677 |
+
{"name": "food processor", "id": 1001, "trainId": 610},
|
| 678 |
+
{"name": "post box", "id": 1958, "trainId": 611},
|
| 679 |
+
{"name": "lid", "id": 1441, "trainId": 612},
|
| 680 |
+
{"name": "drum", "id": 809, "trainId": 613},
|
| 681 |
+
{"name": "blender", "id": 210, "trainId": 614},
|
| 682 |
+
{"name": "cave entrance", "id": 435, "trainId": 615},
|
| 683 |
+
{"name": "dental chair", "id": 718, "trainId": 616},
|
| 684 |
+
{"name": "obelisk", "id": 1674, "trainId": 617},
|
| 685 |
+
{"name": "canoe", "id": 388, "trainId": 618},
|
| 686 |
+
{"name": "mobile", "id": 1572, "trainId": 619},
|
| 687 |
+
{"name": "monitors", "id": 1584, "trainId": 620},
|
| 688 |
+
{"name": "pool ball", "id": 1944, "trainId": 621},
|
| 689 |
+
{"name": "cue rack", "id": 674, "trainId": 622},
|
| 690 |
+
{"name": "baggage carts", "id": 99, "trainId": 623},
|
| 691 |
+
{"name": "shore", "id": 2352, "trainId": 624},
|
| 692 |
+
{"name": "fork", "id": 1019, "trainId": 625},
|
| 693 |
+
{"name": "paper filer", "id": 1763, "trainId": 626},
|
| 694 |
+
{"name": "bicycle rack", "id": 185, "trainId": 627},
|
| 695 |
+
{"name": "coat rack", "id": 554, "trainId": 628},
|
| 696 |
+
{"name": "garland", "id": 1066, "trainId": 629},
|
| 697 |
+
{"name": "sports bag", "id": 2508, "trainId": 630},
|
| 698 |
+
{"name": "fish tank", "id": 951, "trainId": 631},
|
| 699 |
+
{"name": "towel dispenser", "id": 2822, "trainId": 632},
|
| 700 |
+
{"name": "carriage", "id": 415, "trainId": 633},
|
| 701 |
+
{"name": "brochure", "id": 297, "trainId": 634},
|
| 702 |
+
{"name": "plaque", "id": 1914, "trainId": 635},
|
| 703 |
+
{"name": "stringer", "id": 2619, "trainId": 636},
|
| 704 |
+
{"name": "iron", "id": 1338, "trainId": 637},
|
| 705 |
+
{"name": "spoon", "id": 2505, "trainId": 638},
|
| 706 |
+
{"name": "flag pole", "id": 955, "trainId": 639},
|
| 707 |
+
{"name": "toilet brush", "id": 2786, "trainId": 640},
|
| 708 |
+
{"name": "book stand", "id": 238, "trainId": 641},
|
| 709 |
+
{"name": "water faucet, water tap, tap, hydrant", "id": 3000, "trainId": 642},
|
| 710 |
+
{"name": "ticket office", "id": 2763, "trainId": 643},
|
| 711 |
+
{"name": "broom", "id": 299, "trainId": 644},
|
| 712 |
+
{"name": "dvd", "id": 822, "trainId": 645},
|
| 713 |
+
{"name": "ice bucket", "id": 1288, "trainId": 646},
|
| 714 |
+
{"name": "carapace, shell, cuticle, shield", "id": 3101, "trainId": 647},
|
| 715 |
+
{"name": "tureen", "id": 2894, "trainId": 648},
|
| 716 |
+
{"name": "folders", "id": 992, "trainId": 649},
|
| 717 |
+
{"name": "chess", "id": 489, "trainId": 650},
|
| 718 |
+
{"name": "root", "id": 2157, "trainId": 651},
|
| 719 |
+
{"name": "sewing machine", "id": 2309, "trainId": 652},
|
| 720 |
+
{"name": "model", "id": 1576, "trainId": 653},
|
| 721 |
+
{"name": "pen", "id": 1810, "trainId": 654},
|
| 722 |
+
{"name": "violin", "id": 2964, "trainId": 655},
|
| 723 |
+
{"name": "sweatshirt", "id": 2662, "trainId": 656},
|
| 724 |
+
{"name": "recycling materials", "id": 2087, "trainId": 657},
|
| 725 |
+
{"name": "mitten", "id": 1569, "trainId": 658},
|
| 726 |
+
{"name": "chopping board, cutting board", "id": 503, "trainId": 659},
|
| 727 |
+
{"name": "mask", "id": 1505, "trainId": 660},
|
| 728 |
+
{"name": "log", "id": 1468, "trainId": 661},
|
| 729 |
+
{"name": "mouse, computer mouse", "id": 1613, "trainId": 662},
|
| 730 |
+
{"name": "grill", "id": 1138, "trainId": 663},
|
| 731 |
+
{"name": "hole", "id": 1256, "trainId": 664},
|
| 732 |
+
{"name": "target", "id": 2715, "trainId": 665},
|
| 733 |
+
{"name": "trash bag", "id": 2846, "trainId": 666},
|
| 734 |
+
{"name": "chalk", "id": 477, "trainId": 667},
|
| 735 |
+
{"name": "sticks", "id": 2576, "trainId": 668},
|
| 736 |
+
{"name": "balloon", "id": 108, "trainId": 669},
|
| 737 |
+
{"name": "score", "id": 2245, "trainId": 670},
|
| 738 |
+
{"name": "hair spray", "id": 1162, "trainId": 671},
|
| 739 |
+
{"name": "roll", "id": 2149, "trainId": 672},
|
| 740 |
+
{"name": "runner", "id": 2183, "trainId": 673},
|
| 741 |
+
{"name": "engine", "id": 858, "trainId": 674},
|
| 742 |
+
{"name": "inflatable glove", "id": 1324, "trainId": 675},
|
| 743 |
+
{"name": "games", "id": 1055, "trainId": 676},
|
| 744 |
+
{"name": "pallets", "id": 1741, "trainId": 677},
|
| 745 |
+
{"name": "baskets", "id": 149, "trainId": 678},
|
| 746 |
+
{"name": "coop", "id": 615, "trainId": 679},
|
| 747 |
+
{"name": "dvd player", "id": 825, "trainId": 680},
|
| 748 |
+
{"name": "rocking horse", "id": 2143, "trainId": 681},
|
| 749 |
+
{"name": "buckets", "id": 304, "trainId": 682},
|
| 750 |
+
{"name": "bread rolls", "id": 283, "trainId": 683},
|
| 751 |
+
{"name": "shawl", "id": 2322, "trainId": 684},
|
| 752 |
+
{"name": "watering can", "id": 3017, "trainId": 685},
|
| 753 |
+
{"name": "spotlights", "id": 2510, "trainId": 686},
|
| 754 |
+
{"name": "post-it", "id": 1960, "trainId": 687},
|
| 755 |
+
{"name": "bowls", "id": 265, "trainId": 688},
|
| 756 |
+
{"name": "security camera", "id": 2282, "trainId": 689},
|
| 757 |
+
{"name": "runner cloth", "id": 2184, "trainId": 690},
|
| 758 |
+
{"name": "lock", "id": 1461, "trainId": 691},
|
| 759 |
+
{"name": "alarm, warning device, alarm system", "id": 3113, "trainId": 692},
|
| 760 |
+
{"name": "side", "id": 2372, "trainId": 693},
|
| 761 |
+
{"name": "roulette", "id": 2166, "trainId": 694},
|
| 762 |
+
{"name": "bone", "id": 232, "trainId": 695},
|
| 763 |
+
{"name": "cutlery", "id": 693, "trainId": 696},
|
| 764 |
+
{"name": "pool balls", "id": 1945, "trainId": 697},
|
| 765 |
+
{"name": "wheels", "id": 3039, "trainId": 698},
|
| 766 |
+
{"name": "spice rack", "id": 2494, "trainId": 699},
|
| 767 |
+
{"name": "plant pots", "id": 1908, "trainId": 700},
|
| 768 |
+
{"name": "towel ring", "id": 2827, "trainId": 701},
|
| 769 |
+
{"name": "bread box", "id": 280, "trainId": 702},
|
| 770 |
+
{"name": "video", "id": 2950, "trainId": 703},
|
| 771 |
+
{"name": "funfair", "id": 1044, "trainId": 704},
|
| 772 |
+
{"name": "breads", "id": 288, "trainId": 705},
|
| 773 |
+
{"name": "tripod", "id": 2863, "trainId": 706},
|
| 774 |
+
{"name": "ironing board", "id": 1342, "trainId": 707},
|
| 775 |
+
{"name": "skimmer", "id": 2409, "trainId": 708},
|
| 776 |
+
{"name": "hollow", "id": 1258, "trainId": 709},
|
| 777 |
+
{"name": "scratching post", "id": 2249, "trainId": 710},
|
| 778 |
+
{"name": "tricycle", "id": 2862, "trainId": 711},
|
| 779 |
+
{"name": "file box", "id": 920, "trainId": 712},
|
| 780 |
+
{"name": "mountain pass", "id": 1607, "trainId": 713},
|
| 781 |
+
{"name": "tombstones", "id": 2802, "trainId": 714},
|
| 782 |
+
{"name": "cooker", "id": 610, "trainId": 715},
|
| 783 |
+
{"name": "card game, cards", "id": 3129, "trainId": 716},
|
| 784 |
+
{"name": "golf bag", "id": 1108, "trainId": 717},
|
| 785 |
+
{"name": "towel paper", "id": 2823, "trainId": 718},
|
| 786 |
+
{"name": "chaise lounge", "id": 476, "trainId": 719},
|
| 787 |
+
{"name": "sun", "id": 2641, "trainId": 720},
|
| 788 |
+
{"name": "toilet paper holder", "id": 2788, "trainId": 721},
|
| 789 |
+
{"name": "rake", "id": 2070, "trainId": 722},
|
| 790 |
+
{"name": "key", "id": 1368, "trainId": 723},
|
| 791 |
+
{"name": "umbrella stand", "id": 2903, "trainId": 724},
|
| 792 |
+
{"name": "dartboard", "id": 699, "trainId": 725},
|
| 793 |
+
{"name": "transformer", "id": 2844, "trainId": 726},
|
| 794 |
+
{"name": "fireplace utensils", "id": 942, "trainId": 727},
|
| 795 |
+
{"name": "sweatshirts", "id": 2663, "trainId": 728},
|
| 796 |
+
{
|
| 797 |
+
"name": "cellular telephone, cellular phone, cellphone, cell, mobile phone",
|
| 798 |
+
"id": 457,
|
| 799 |
+
"trainId": 729,
|
| 800 |
+
},
|
| 801 |
+
{"name": "tallboy", "id": 2701, "trainId": 730},
|
| 802 |
+
{"name": "stapler", "id": 2540, "trainId": 731},
|
| 803 |
+
{"name": "sauna", "id": 2231, "trainId": 732},
|
| 804 |
+
{"name": "test tube", "id": 2746, "trainId": 733},
|
| 805 |
+
{"name": "palette", "id": 1738, "trainId": 734},
|
| 806 |
+
{"name": "shopping carts", "id": 2350, "trainId": 735},
|
| 807 |
+
{"name": "tools", "id": 2808, "trainId": 736},
|
| 808 |
+
{"name": "push button, push, button", "id": 2025, "trainId": 737},
|
| 809 |
+
{"name": "star", "id": 2541, "trainId": 738},
|
| 810 |
+
{"name": "roof rack", "id": 2156, "trainId": 739},
|
| 811 |
+
{"name": "barbed wire", "id": 126, "trainId": 740},
|
| 812 |
+
{"name": "spray", "id": 2512, "trainId": 741},
|
| 813 |
+
{"name": "ear", "id": 831, "trainId": 742},
|
| 814 |
+
{"name": "sponge", "id": 2503, "trainId": 743},
|
| 815 |
+
{"name": "racket", "id": 2039, "trainId": 744},
|
| 816 |
+
{"name": "tins", "id": 2774, "trainId": 745},
|
| 817 |
+
{"name": "eyeglasses", "id": 886, "trainId": 746},
|
| 818 |
+
{"name": "file", "id": 919, "trainId": 747},
|
| 819 |
+
{"name": "scarfs", "id": 2240, "trainId": 748},
|
| 820 |
+
{"name": "sugar bowl", "id": 2636, "trainId": 749},
|
| 821 |
+
{"name": "flip flop", "id": 963, "trainId": 750},
|
| 822 |
+
{"name": "headstones", "id": 1218, "trainId": 751},
|
| 823 |
+
{"name": "laptop bag", "id": 1406, "trainId": 752},
|
| 824 |
+
{"name": "leash", "id": 1420, "trainId": 753},
|
| 825 |
+
{"name": "climbing frame", "id": 526, "trainId": 754},
|
| 826 |
+
{"name": "suit hanger", "id": 2639, "trainId": 755},
|
| 827 |
+
{"name": "floor spotlight", "id": 975, "trainId": 756},
|
| 828 |
+
{"name": "plate rack", "id": 1921, "trainId": 757},
|
| 829 |
+
{"name": "sewer", "id": 2305, "trainId": 758},
|
| 830 |
+
{"name": "hard drive", "id": 1193, "trainId": 759},
|
| 831 |
+
{"name": "sprinkler", "id": 2517, "trainId": 760},
|
| 832 |
+
{"name": "tools box", "id": 2809, "trainId": 761},
|
| 833 |
+
{"name": "necklace", "id": 1647, "trainId": 762},
|
| 834 |
+
{"name": "bulbs", "id": 314, "trainId": 763},
|
| 835 |
+
{"name": "steel industry", "id": 2560, "trainId": 764},
|
| 836 |
+
{"name": "club", "id": 545, "trainId": 765},
|
| 837 |
+
{"name": "jack", "id": 1345, "trainId": 766},
|
| 838 |
+
{"name": "door bars", "id": 775, "trainId": 767},
|
| 839 |
+
{
|
| 840 |
+
"name": "control panel, instrument panel, control board, board, panel",
|
| 841 |
+
"id": 603,
|
| 842 |
+
"trainId": 768,
|
| 843 |
+
},
|
| 844 |
+
{"name": "hairbrush", "id": 1163, "trainId": 769},
|
| 845 |
+
{"name": "napkin holder", "id": 1641, "trainId": 770},
|
| 846 |
+
{"name": "office", "id": 1678, "trainId": 771},
|
| 847 |
+
{"name": "smoke detector", "id": 2450, "trainId": 772},
|
| 848 |
+
{"name": "utensils", "id": 2915, "trainId": 773},
|
| 849 |
+
{"name": "apron", "id": 42, "trainId": 774},
|
| 850 |
+
{"name": "scissors", "id": 2242, "trainId": 775},
|
| 851 |
+
{"name": "terminal", "id": 2741, "trainId": 776},
|
| 852 |
+
{"name": "grinder", "id": 1143, "trainId": 777},
|
| 853 |
+
{"name": "entry phone", "id": 862, "trainId": 778},
|
| 854 |
+
{"name": "newspaper stand", "id": 1654, "trainId": 779},
|
| 855 |
+
{"name": "pepper shaker", "id": 1826, "trainId": 780},
|
| 856 |
+
{"name": "onions", "id": 1689, "trainId": 781},
|
| 857 |
+
{
|
| 858 |
+
"name": "central processing unit, cpu, c p u , central processor, processor, mainframe",
|
| 859 |
+
"id": 3124,
|
| 860 |
+
"trainId": 782,
|
| 861 |
+
},
|
| 862 |
+
{"name": "tape", "id": 2710, "trainId": 783},
|
| 863 |
+
{"name": "bat", "id": 152, "trainId": 784},
|
| 864 |
+
{"name": "coaster", "id": 549, "trainId": 785},
|
| 865 |
+
{"name": "calculator", "id": 360, "trainId": 786},
|
| 866 |
+
{"name": "potatoes", "id": 1982, "trainId": 787},
|
| 867 |
+
{"name": "luggage rack", "id": 1478, "trainId": 788},
|
| 868 |
+
{"name": "salt", "id": 2203, "trainId": 789},
|
| 869 |
+
{"name": "street number", "id": 2612, "trainId": 790},
|
| 870 |
+
{"name": "viewpoint", "id": 2956, "trainId": 791},
|
| 871 |
+
{"name": "sword", "id": 2681, "trainId": 792},
|
| 872 |
+
{"name": "cd", "id": 437, "trainId": 793},
|
| 873 |
+
{"name": "rowing machine", "id": 2171, "trainId": 794},
|
| 874 |
+
{"name": "plug", "id": 1933, "trainId": 795},
|
| 875 |
+
{"name": "andiron, firedog, dog, dog-iron", "id": 3110, "trainId": 796},
|
| 876 |
+
{"name": "pepper", "id": 1824, "trainId": 797},
|
| 877 |
+
{"name": "tongs", "id": 2803, "trainId": 798},
|
| 878 |
+
{"name": "bonfire", "id": 234, "trainId": 799},
|
| 879 |
+
{"name": "dog dish", "id": 764, "trainId": 800},
|
| 880 |
+
{"name": "belt", "id": 177, "trainId": 801},
|
| 881 |
+
{"name": "dumbbells", "id": 817, "trainId": 802},
|
| 882 |
+
{"name": "videocassette recorder, vcr", "id": 3145, "trainId": 803},
|
| 883 |
+
{"name": "hook", "id": 1262, "trainId": 804},
|
| 884 |
+
{"name": "envelopes", "id": 864, "trainId": 805},
|
| 885 |
+
{"name": "shower faucet", "id": 2359, "trainId": 806},
|
| 886 |
+
{"name": "watch", "id": 2992, "trainId": 807},
|
| 887 |
+
{"name": "padlock", "id": 1725, "trainId": 808},
|
| 888 |
+
{"name": "swimming pool ladder", "id": 2667, "trainId": 809},
|
| 889 |
+
{"name": "spanners", "id": 2484, "trainId": 810},
|
| 890 |
+
{"name": "gravy boat", "id": 1133, "trainId": 811},
|
| 891 |
+
{"name": "notice board", "id": 1667, "trainId": 812},
|
| 892 |
+
{"name": "trash bags", "id": 2847, "trainId": 813},
|
| 893 |
+
{"name": "fire alarm", "id": 932, "trainId": 814},
|
| 894 |
+
{"name": "ladle", "id": 1392, "trainId": 815},
|
| 895 |
+
{"name": "stethoscope", "id": 2573, "trainId": 816},
|
| 896 |
+
{"name": "rocket", "id": 2140, "trainId": 817},
|
| 897 |
+
{"name": "funnel", "id": 1046, "trainId": 818},
|
| 898 |
+
{"name": "bowling pins", "id": 264, "trainId": 819},
|
| 899 |
+
{"name": "valve", "id": 2927, "trainId": 820},
|
| 900 |
+
{"name": "thermometer", "id": 2752, "trainId": 821},
|
| 901 |
+
{"name": "cups", "id": 679, "trainId": 822},
|
| 902 |
+
{"name": "spice jar", "id": 2493, "trainId": 823},
|
| 903 |
+
{"name": "night light", "id": 1658, "trainId": 824},
|
| 904 |
+
{"name": "soaps", "id": 2466, "trainId": 825},
|
| 905 |
+
{"name": "games table", "id": 1057, "trainId": 826},
|
| 906 |
+
{"name": "slotted spoon", "id": 2444, "trainId": 827},
|
| 907 |
+
{"name": "reel", "id": 2093, "trainId": 828},
|
| 908 |
+
{"name": "scourer", "id": 2248, "trainId": 829},
|
| 909 |
+
{"name": "sleeping robe", "id": 2432, "trainId": 830},
|
| 910 |
+
{"name": "desk mat", "id": 726, "trainId": 831},
|
| 911 |
+
{"name": "dumbbell", "id": 816, "trainId": 832},
|
| 912 |
+
{"name": "hammer", "id": 1171, "trainId": 833},
|
| 913 |
+
{"name": "tie", "id": 2766, "trainId": 834},
|
| 914 |
+
{"name": "typewriter", "id": 2900, "trainId": 835},
|
| 915 |
+
{"name": "shaker", "id": 2313, "trainId": 836},
|
| 916 |
+
{"name": "cheese dish", "id": 488, "trainId": 837},
|
| 917 |
+
{"name": "sea star", "id": 2265, "trainId": 838},
|
| 918 |
+
{"name": "racquet", "id": 2043, "trainId": 839},
|
| 919 |
+
{"name": "butane gas cylinder", "id": 332, "trainId": 840},
|
| 920 |
+
{"name": "paper weight", "id": 1771, "trainId": 841},
|
| 921 |
+
{"name": "shaving brush", "id": 2320, "trainId": 842},
|
| 922 |
+
{"name": "sunglasses", "id": 2646, "trainId": 843},
|
| 923 |
+
{"name": "gear shift", "id": 1089, "trainId": 844},
|
| 924 |
+
{"name": "towel rail", "id": 2826, "trainId": 845},
|
| 925 |
+
{"name": "adding machine, totalizer, totaliser", "id": 3148, "trainId": 846},
|
| 926 |
+
]
|
| 927 |
+
|
| 928 |
+
|
| 929 |
+
def loadAde20K(file):
|
| 930 |
+
fileseg = file.replace(".jpg", "_seg.png")
|
| 931 |
+
with Image.open(fileseg) as io:
|
| 932 |
+
seg = np.array(io)
|
| 933 |
+
|
| 934 |
+
R = seg[:, :, 0]
|
| 935 |
+
G = seg[:, :, 1]
|
| 936 |
+
ObjectClassMasks = (R / 10).astype(np.int32) * 256 + (G.astype(np.int32))
|
| 937 |
+
|
| 938 |
+
return {"img_name": file, "segm_name": fileseg, "class_mask": ObjectClassMasks}
|
| 939 |
+
|
| 940 |
+
|
| 941 |
+
if __name__ == "__main__":
|
| 942 |
+
dataset_dir = Path(os.getenv("DETECTRON2_DATASETS", "datasets"))
|
| 943 |
+
index_file = dataset_dir / "ade/ADE20K_2021_17_01" / "index_ade20k.pkl"
|
| 944 |
+
with open(index_file, "rb") as f:
|
| 945 |
+
index_ade20k = pkl.load(f)
|
| 946 |
+
|
| 947 |
+
id_map = {}
|
| 948 |
+
for cat in ADE20K_SEM_SEG_FULL_CATEGORIES:
|
| 949 |
+
id_map[cat["id"]] = cat["trainId"]
|
| 950 |
+
|
| 951 |
+
# make output dir
|
| 952 |
+
for name in ["training", "validation"]:
|
| 953 |
+
image_dir = dataset_dir / "ade/ADE20K_2021_17_01" / "images_detectron2" / name
|
| 954 |
+
image_dir.mkdir(parents=True, exist_ok=True)
|
| 955 |
+
annotation_dir = dataset_dir / "ade/ADE20K_2021_17_01" / "annotations_detectron2" / name
|
| 956 |
+
annotation_dir.mkdir(parents=True, exist_ok=True)
|
| 957 |
+
|
| 958 |
+
# process image and gt
|
| 959 |
+
for folder_name, file_name in tqdm.tqdm(
|
| 960 |
+
zip(index_ade20k["folder"], index_ade20k["filename"]),
|
| 961 |
+
total=len(index_ade20k["filename"]),
|
| 962 |
+
):
|
| 963 |
+
split = "validation" if file_name.split("_")[1] == "val" else "training"
|
| 964 |
+
info = loadAde20K(str(dataset_dir / "ade" / folder_name / file_name))
|
| 965 |
+
|
| 966 |
+
# resize image and label
|
| 967 |
+
img = np.asarray(Image.open(info["img_name"]))
|
| 968 |
+
lab = np.asarray(info["class_mask"])
|
| 969 |
+
|
| 970 |
+
h, w = img.shape[0], img.shape[1]
|
| 971 |
+
max_size = 512
|
| 972 |
+
resize = True
|
| 973 |
+
if w >= h > max_size:
|
| 974 |
+
h_new, w_new = max_size, round(w / float(h) * max_size)
|
| 975 |
+
elif h >= w > max_size:
|
| 976 |
+
h_new, w_new = round(h / float(w) * max_size), max_size
|
| 977 |
+
else:
|
| 978 |
+
resize = False
|
| 979 |
+
|
| 980 |
+
if resize:
|
| 981 |
+
img = cv2.resize(img, (w_new, h_new), interpolation=cv2.INTER_LINEAR)
|
| 982 |
+
lab = cv2.resize(lab, (w_new, h_new), interpolation=cv2.INTER_NEAREST)
|
| 983 |
+
|
| 984 |
+
assert img.dtype == np.uint8
|
| 985 |
+
assert lab.dtype == np.int32
|
| 986 |
+
|
| 987 |
+
# apply label conversion and save into uint16 images
|
| 988 |
+
output = np.zeros_like(lab, dtype=np.uint16) + 65535
|
| 989 |
+
for obj_id in np.unique(lab):
|
| 990 |
+
if obj_id in id_map:
|
| 991 |
+
output[lab == obj_id] = id_map[obj_id]
|
| 992 |
+
|
| 993 |
+
output_img = dataset_dir / "ade/ADE20K_2021_17_01" / "images_detectron2" / split / file_name
|
| 994 |
+
output_lab = (
|
| 995 |
+
dataset_dir
|
| 996 |
+
/ "ade/ADE20K_2021_17_01"
|
| 997 |
+
/ "annotations_detectron2"
|
| 998 |
+
/ split
|
| 999 |
+
/ file_name.replace(".jpg", ".tif")
|
| 1000 |
+
)
|
| 1001 |
+
Image.fromarray(img).save(output_img)
|
| 1002 |
+
|
| 1003 |
+
assert output.dtype == np.uint16
|
| 1004 |
+
Image.fromarray(output).save(output_lab)
|
datasets/prepare_ade20k_ins_seg.py
ADDED
|
@@ -0,0 +1,111 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
+
# -*- coding: utf-8 -*-
|
| 3 |
+
import glob
|
| 4 |
+
import json
|
| 5 |
+
import os
|
| 6 |
+
from collections import Counter
|
| 7 |
+
|
| 8 |
+
import numpy as np
|
| 9 |
+
import tqdm
|
| 10 |
+
from panopticapi.utils import IdGenerator, save_json
|
| 11 |
+
from PIL import Image
|
| 12 |
+
import pycocotools.mask as mask_util
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
if __name__ == "__main__":
|
| 16 |
+
dataset_dir = os.getenv("DETECTRON2_DATASETS", "datasets")
|
| 17 |
+
|
| 18 |
+
for name, dirname in [("train", "training"), ("val", "validation")]:
|
| 19 |
+
image_dir = os.path.join(dataset_dir, f"ADEChallengeData2016/images/{dirname}/")
|
| 20 |
+
instance_dir = os.path.join(
|
| 21 |
+
dataset_dir, f"ADEChallengeData2016/annotations_instance/{dirname}/"
|
| 22 |
+
)
|
| 23 |
+
|
| 24 |
+
# img_id = 0
|
| 25 |
+
ann_id = 1
|
| 26 |
+
|
| 27 |
+
# json
|
| 28 |
+
out_file = os.path.join(dataset_dir, f"ADEChallengeData2016/ade20k_instance_{name}.json")
|
| 29 |
+
|
| 30 |
+
# json config
|
| 31 |
+
instance_config_file = "datasets/ade20k_instance_imgCatIds.json"
|
| 32 |
+
with open(instance_config_file) as f:
|
| 33 |
+
category_dict = json.load(f)["categories"]
|
| 34 |
+
|
| 35 |
+
# load catid mapping
|
| 36 |
+
# it is important to share category id for both instance and panoptic annotations
|
| 37 |
+
mapping_file = "datasets/ade20k_instance_catid_mapping.txt"
|
| 38 |
+
with open(mapping_file) as f:
|
| 39 |
+
map_id = {}
|
| 40 |
+
for i, line in enumerate(f.readlines()):
|
| 41 |
+
if i == 0:
|
| 42 |
+
continue
|
| 43 |
+
ins_id, sem_id, _ = line.strip().split()
|
| 44 |
+
# shift id by 1 because we want it to start from 0!
|
| 45 |
+
# ignore_label becomes 255
|
| 46 |
+
map_id[int(ins_id)] = int(sem_id) - 1
|
| 47 |
+
|
| 48 |
+
for cat in category_dict:
|
| 49 |
+
cat["id"] = map_id[cat["id"]]
|
| 50 |
+
|
| 51 |
+
filenames = sorted(glob.glob(os.path.join(image_dir, "*.jpg")))
|
| 52 |
+
|
| 53 |
+
ann_dict = {}
|
| 54 |
+
images = []
|
| 55 |
+
annotations = []
|
| 56 |
+
|
| 57 |
+
for idx, filename in enumerate(tqdm.tqdm(filenames)):
|
| 58 |
+
image = {}
|
| 59 |
+
image_id = os.path.basename(filename).split(".")[0]
|
| 60 |
+
|
| 61 |
+
image["id"] = image_id
|
| 62 |
+
image["file_name"] = os.path.basename(filename)
|
| 63 |
+
|
| 64 |
+
original_format = np.array(Image.open(filename))
|
| 65 |
+
image["width"] = original_format.shape[1]
|
| 66 |
+
image["height"] = original_format.shape[0]
|
| 67 |
+
|
| 68 |
+
images.append(image)
|
| 69 |
+
|
| 70 |
+
filename_instance = os.path.join(instance_dir, image_id + ".png")
|
| 71 |
+
ins_seg = np.asarray(Image.open(filename_instance))
|
| 72 |
+
assert ins_seg.dtype == np.uint8
|
| 73 |
+
|
| 74 |
+
instance_cat_ids = ins_seg[..., 0]
|
| 75 |
+
# instance id starts from 1!
|
| 76 |
+
# because 0 is reserved as VOID label
|
| 77 |
+
instance_ins_ids = ins_seg[..., 1]
|
| 78 |
+
|
| 79 |
+
# process things
|
| 80 |
+
for thing_id in np.unique(instance_ins_ids):
|
| 81 |
+
if thing_id == 0:
|
| 82 |
+
continue
|
| 83 |
+
mask = instance_ins_ids == thing_id
|
| 84 |
+
instance_cat_id = np.unique(instance_cat_ids[mask])
|
| 85 |
+
assert len(instance_cat_id) == 1
|
| 86 |
+
|
| 87 |
+
anno = {}
|
| 88 |
+
anno['id'] = ann_id
|
| 89 |
+
ann_id += 1
|
| 90 |
+
anno['image_id'] = image['id']
|
| 91 |
+
anno["iscrowd"] = int(0)
|
| 92 |
+
anno["category_id"] = int(map_id[instance_cat_id[0]])
|
| 93 |
+
|
| 94 |
+
inds = np.nonzero(mask)
|
| 95 |
+
ymin, ymax = inds[0].min(), inds[0].max()
|
| 96 |
+
xmin, xmax = inds[1].min(), inds[1].max()
|
| 97 |
+
anno["bbox"] = [int(xmin), int(ymin), int(xmax - xmin + 1), int(ymax - ymin + 1)]
|
| 98 |
+
# if xmax <= xmin or ymax <= ymin:
|
| 99 |
+
# continue
|
| 100 |
+
rle = mask_util.encode(np.array(mask[:, :, None], order="F", dtype="uint8"))[0]
|
| 101 |
+
rle["counts"] = rle["counts"].decode("utf-8")
|
| 102 |
+
anno["segmentation"] = rle
|
| 103 |
+
anno["area"] = int(mask_util.area(rle))
|
| 104 |
+
annotations.append(anno)
|
| 105 |
+
|
| 106 |
+
# save this
|
| 107 |
+
ann_dict['images'] = images
|
| 108 |
+
ann_dict['categories'] = category_dict
|
| 109 |
+
ann_dict['annotations'] = annotations
|
| 110 |
+
|
| 111 |
+
save_json(ann_dict, out_file)
|
datasets/prepare_ade20k_pan_seg.py
ADDED
|
@@ -0,0 +1,499 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
+
# -*- coding: utf-8 -*-
|
| 3 |
+
import glob
|
| 4 |
+
import json
|
| 5 |
+
import os
|
| 6 |
+
from collections import Counter
|
| 7 |
+
|
| 8 |
+
import numpy as np
|
| 9 |
+
import tqdm
|
| 10 |
+
from panopticapi.utils import IdGenerator, save_json
|
| 11 |
+
from PIL import Image
|
| 12 |
+
|
| 13 |
+
ADE20K_SEM_SEG_CATEGORIES = [
|
| 14 |
+
"wall",
|
| 15 |
+
"building",
|
| 16 |
+
"sky",
|
| 17 |
+
"floor",
|
| 18 |
+
"tree",
|
| 19 |
+
"ceiling",
|
| 20 |
+
"road, route",
|
| 21 |
+
"bed",
|
| 22 |
+
"window ",
|
| 23 |
+
"grass",
|
| 24 |
+
"cabinet",
|
| 25 |
+
"sidewalk, pavement",
|
| 26 |
+
"person",
|
| 27 |
+
"earth, ground",
|
| 28 |
+
"door",
|
| 29 |
+
"table",
|
| 30 |
+
"mountain, mount",
|
| 31 |
+
"plant",
|
| 32 |
+
"curtain",
|
| 33 |
+
"chair",
|
| 34 |
+
"car",
|
| 35 |
+
"water",
|
| 36 |
+
"painting, picture",
|
| 37 |
+
"sofa",
|
| 38 |
+
"shelf",
|
| 39 |
+
"house",
|
| 40 |
+
"sea",
|
| 41 |
+
"mirror",
|
| 42 |
+
"rug",
|
| 43 |
+
"field",
|
| 44 |
+
"armchair",
|
| 45 |
+
"seat",
|
| 46 |
+
"fence",
|
| 47 |
+
"desk",
|
| 48 |
+
"rock, stone",
|
| 49 |
+
"wardrobe, closet, press",
|
| 50 |
+
"lamp",
|
| 51 |
+
"tub",
|
| 52 |
+
"rail",
|
| 53 |
+
"cushion",
|
| 54 |
+
"base, pedestal, stand",
|
| 55 |
+
"box",
|
| 56 |
+
"column, pillar",
|
| 57 |
+
"signboard, sign",
|
| 58 |
+
"chest of drawers, chest, bureau, dresser",
|
| 59 |
+
"counter",
|
| 60 |
+
"sand",
|
| 61 |
+
"sink",
|
| 62 |
+
"skyscraper",
|
| 63 |
+
"fireplace",
|
| 64 |
+
"refrigerator, icebox",
|
| 65 |
+
"grandstand, covered stand",
|
| 66 |
+
"path",
|
| 67 |
+
"stairs",
|
| 68 |
+
"runway",
|
| 69 |
+
"case, display case, showcase, vitrine",
|
| 70 |
+
"pool table, billiard table, snooker table",
|
| 71 |
+
"pillow",
|
| 72 |
+
"screen door, screen",
|
| 73 |
+
"stairway, staircase",
|
| 74 |
+
"river",
|
| 75 |
+
"bridge, span",
|
| 76 |
+
"bookcase",
|
| 77 |
+
"blind, screen",
|
| 78 |
+
"coffee table",
|
| 79 |
+
"toilet, can, commode, crapper, pot, potty, stool, throne",
|
| 80 |
+
"flower",
|
| 81 |
+
"book",
|
| 82 |
+
"hill",
|
| 83 |
+
"bench",
|
| 84 |
+
"countertop",
|
| 85 |
+
"stove",
|
| 86 |
+
"palm, palm tree",
|
| 87 |
+
"kitchen island",
|
| 88 |
+
"computer",
|
| 89 |
+
"swivel chair",
|
| 90 |
+
"boat",
|
| 91 |
+
"bar",
|
| 92 |
+
"arcade machine",
|
| 93 |
+
"hovel, hut, hutch, shack, shanty",
|
| 94 |
+
"bus",
|
| 95 |
+
"towel",
|
| 96 |
+
"light",
|
| 97 |
+
"truck",
|
| 98 |
+
"tower",
|
| 99 |
+
"chandelier",
|
| 100 |
+
"awning, sunshade, sunblind",
|
| 101 |
+
"street lamp",
|
| 102 |
+
"booth",
|
| 103 |
+
"tv",
|
| 104 |
+
"plane",
|
| 105 |
+
"dirt track",
|
| 106 |
+
"clothes",
|
| 107 |
+
"pole",
|
| 108 |
+
"land, ground, soil",
|
| 109 |
+
"bannister, banister, balustrade, balusters, handrail",
|
| 110 |
+
"escalator, moving staircase, moving stairway",
|
| 111 |
+
"ottoman, pouf, pouffe, puff, hassock",
|
| 112 |
+
"bottle",
|
| 113 |
+
"buffet, counter, sideboard",
|
| 114 |
+
"poster, posting, placard, notice, bill, card",
|
| 115 |
+
"stage",
|
| 116 |
+
"van",
|
| 117 |
+
"ship",
|
| 118 |
+
"fountain",
|
| 119 |
+
"conveyer belt, conveyor belt, conveyer, conveyor, transporter",
|
| 120 |
+
"canopy",
|
| 121 |
+
"washer, automatic washer, washing machine",
|
| 122 |
+
"plaything, toy",
|
| 123 |
+
"pool",
|
| 124 |
+
"stool",
|
| 125 |
+
"barrel, cask",
|
| 126 |
+
"basket, handbasket",
|
| 127 |
+
"falls",
|
| 128 |
+
"tent",
|
| 129 |
+
"bag",
|
| 130 |
+
"minibike, motorbike",
|
| 131 |
+
"cradle",
|
| 132 |
+
"oven",
|
| 133 |
+
"ball",
|
| 134 |
+
"food, solid food",
|
| 135 |
+
"step, stair",
|
| 136 |
+
"tank, storage tank",
|
| 137 |
+
"trade name",
|
| 138 |
+
"microwave",
|
| 139 |
+
"pot",
|
| 140 |
+
"animal",
|
| 141 |
+
"bicycle",
|
| 142 |
+
"lake",
|
| 143 |
+
"dishwasher",
|
| 144 |
+
"screen",
|
| 145 |
+
"blanket, cover",
|
| 146 |
+
"sculpture",
|
| 147 |
+
"hood, exhaust hood",
|
| 148 |
+
"sconce",
|
| 149 |
+
"vase",
|
| 150 |
+
"traffic light",
|
| 151 |
+
"tray",
|
| 152 |
+
"trash can",
|
| 153 |
+
"fan",
|
| 154 |
+
"pier",
|
| 155 |
+
"crt screen",
|
| 156 |
+
"plate",
|
| 157 |
+
"monitor",
|
| 158 |
+
"bulletin board",
|
| 159 |
+
"shower",
|
| 160 |
+
"radiator",
|
| 161 |
+
"glass, drinking glass",
|
| 162 |
+
"clock",
|
| 163 |
+
"flag", # noqa
|
| 164 |
+
]
|
| 165 |
+
|
| 166 |
+
PALETTE = [
|
| 167 |
+
[120, 120, 120],
|
| 168 |
+
[180, 120, 120],
|
| 169 |
+
[6, 230, 230],
|
| 170 |
+
[80, 50, 50],
|
| 171 |
+
[4, 200, 3],
|
| 172 |
+
[120, 120, 80],
|
| 173 |
+
[140, 140, 140],
|
| 174 |
+
[204, 5, 255],
|
| 175 |
+
[230, 230, 230],
|
| 176 |
+
[4, 250, 7],
|
| 177 |
+
[224, 5, 255],
|
| 178 |
+
[235, 255, 7],
|
| 179 |
+
[150, 5, 61],
|
| 180 |
+
[120, 120, 70],
|
| 181 |
+
[8, 255, 51],
|
| 182 |
+
[255, 6, 82],
|
| 183 |
+
[143, 255, 140],
|
| 184 |
+
[204, 255, 4],
|
| 185 |
+
[255, 51, 7],
|
| 186 |
+
[204, 70, 3],
|
| 187 |
+
[0, 102, 200],
|
| 188 |
+
[61, 230, 250],
|
| 189 |
+
[255, 6, 51],
|
| 190 |
+
[11, 102, 255],
|
| 191 |
+
[255, 7, 71],
|
| 192 |
+
[255, 9, 224],
|
| 193 |
+
[9, 7, 230],
|
| 194 |
+
[220, 220, 220],
|
| 195 |
+
[255, 9, 92],
|
| 196 |
+
[112, 9, 255],
|
| 197 |
+
[8, 255, 214],
|
| 198 |
+
[7, 255, 224],
|
| 199 |
+
[255, 184, 6],
|
| 200 |
+
[10, 255, 71],
|
| 201 |
+
[255, 41, 10],
|
| 202 |
+
[7, 255, 255],
|
| 203 |
+
[224, 255, 8],
|
| 204 |
+
[102, 8, 255],
|
| 205 |
+
[255, 61, 6],
|
| 206 |
+
[255, 194, 7],
|
| 207 |
+
[255, 122, 8],
|
| 208 |
+
[0, 255, 20],
|
| 209 |
+
[255, 8, 41],
|
| 210 |
+
[255, 5, 153],
|
| 211 |
+
[6, 51, 255],
|
| 212 |
+
[235, 12, 255],
|
| 213 |
+
[160, 150, 20],
|
| 214 |
+
[0, 163, 255],
|
| 215 |
+
[140, 140, 200],
|
| 216 |
+
[250, 10, 15],
|
| 217 |
+
[20, 255, 0],
|
| 218 |
+
[31, 255, 0],
|
| 219 |
+
[255, 31, 0],
|
| 220 |
+
[255, 224, 0],
|
| 221 |
+
[153, 255, 0],
|
| 222 |
+
[0, 0, 255],
|
| 223 |
+
[255, 71, 0],
|
| 224 |
+
[0, 235, 255],
|
| 225 |
+
[0, 173, 255],
|
| 226 |
+
[31, 0, 255],
|
| 227 |
+
[11, 200, 200],
|
| 228 |
+
[255, 82, 0],
|
| 229 |
+
[0, 255, 245],
|
| 230 |
+
[0, 61, 255],
|
| 231 |
+
[0, 255, 112],
|
| 232 |
+
[0, 255, 133],
|
| 233 |
+
[255, 0, 0],
|
| 234 |
+
[255, 163, 0],
|
| 235 |
+
[255, 102, 0],
|
| 236 |
+
[194, 255, 0],
|
| 237 |
+
[0, 143, 255],
|
| 238 |
+
[51, 255, 0],
|
| 239 |
+
[0, 82, 255],
|
| 240 |
+
[0, 255, 41],
|
| 241 |
+
[0, 255, 173],
|
| 242 |
+
[10, 0, 255],
|
| 243 |
+
[173, 255, 0],
|
| 244 |
+
[0, 255, 153],
|
| 245 |
+
[255, 92, 0],
|
| 246 |
+
[255, 0, 255],
|
| 247 |
+
[255, 0, 245],
|
| 248 |
+
[255, 0, 102],
|
| 249 |
+
[255, 173, 0],
|
| 250 |
+
[255, 0, 20],
|
| 251 |
+
[255, 184, 184],
|
| 252 |
+
[0, 31, 255],
|
| 253 |
+
[0, 255, 61],
|
| 254 |
+
[0, 71, 255],
|
| 255 |
+
[255, 0, 204],
|
| 256 |
+
[0, 255, 194],
|
| 257 |
+
[0, 255, 82],
|
| 258 |
+
[0, 10, 255],
|
| 259 |
+
[0, 112, 255],
|
| 260 |
+
[51, 0, 255],
|
| 261 |
+
[0, 194, 255],
|
| 262 |
+
[0, 122, 255],
|
| 263 |
+
[0, 255, 163],
|
| 264 |
+
[255, 153, 0],
|
| 265 |
+
[0, 255, 10],
|
| 266 |
+
[255, 112, 0],
|
| 267 |
+
[143, 255, 0],
|
| 268 |
+
[82, 0, 255],
|
| 269 |
+
[163, 255, 0],
|
| 270 |
+
[255, 235, 0],
|
| 271 |
+
[8, 184, 170],
|
| 272 |
+
[133, 0, 255],
|
| 273 |
+
[0, 255, 92],
|
| 274 |
+
[184, 0, 255],
|
| 275 |
+
[255, 0, 31],
|
| 276 |
+
[0, 184, 255],
|
| 277 |
+
[0, 214, 255],
|
| 278 |
+
[255, 0, 112],
|
| 279 |
+
[92, 255, 0],
|
| 280 |
+
[0, 224, 255],
|
| 281 |
+
[112, 224, 255],
|
| 282 |
+
[70, 184, 160],
|
| 283 |
+
[163, 0, 255],
|
| 284 |
+
[153, 0, 255],
|
| 285 |
+
[71, 255, 0],
|
| 286 |
+
[255, 0, 163],
|
| 287 |
+
[255, 204, 0],
|
| 288 |
+
[255, 0, 143],
|
| 289 |
+
[0, 255, 235],
|
| 290 |
+
[133, 255, 0],
|
| 291 |
+
[255, 0, 235],
|
| 292 |
+
[245, 0, 255],
|
| 293 |
+
[255, 0, 122],
|
| 294 |
+
[255, 245, 0],
|
| 295 |
+
[10, 190, 212],
|
| 296 |
+
[214, 255, 0],
|
| 297 |
+
[0, 204, 255],
|
| 298 |
+
[20, 0, 255],
|
| 299 |
+
[255, 255, 0],
|
| 300 |
+
[0, 153, 255],
|
| 301 |
+
[0, 41, 255],
|
| 302 |
+
[0, 255, 204],
|
| 303 |
+
[41, 0, 255],
|
| 304 |
+
[41, 255, 0],
|
| 305 |
+
[173, 0, 255],
|
| 306 |
+
[0, 245, 255],
|
| 307 |
+
[71, 0, 255],
|
| 308 |
+
[122, 0, 255],
|
| 309 |
+
[0, 255, 184],
|
| 310 |
+
[0, 92, 255],
|
| 311 |
+
[184, 255, 0],
|
| 312 |
+
[0, 133, 255],
|
| 313 |
+
[255, 214, 0],
|
| 314 |
+
[25, 194, 194],
|
| 315 |
+
[102, 255, 0],
|
| 316 |
+
[92, 0, 255],
|
| 317 |
+
]
|
| 318 |
+
|
| 319 |
+
|
| 320 |
+
if __name__ == "__main__":
|
| 321 |
+
dataset_dir = os.getenv("DETECTRON2_DATASETS", "datasets")
|
| 322 |
+
|
| 323 |
+
for name, dirname in [("train", "training"), ("val", "validation")]:
|
| 324 |
+
image_dir = os.path.join(dataset_dir, f"ADEChallengeData2016/images/{dirname}/")
|
| 325 |
+
semantic_dir = os.path.join(dataset_dir, f"ADEChallengeData2016/annotations/{dirname}/")
|
| 326 |
+
instance_dir = os.path.join(
|
| 327 |
+
dataset_dir, f"ADEChallengeData2016/annotations_instance/{dirname}/"
|
| 328 |
+
)
|
| 329 |
+
|
| 330 |
+
# folder to store panoptic PNGs
|
| 331 |
+
out_folder = os.path.join(dataset_dir, f"ADEChallengeData2016/ade20k_panoptic_{name}/")
|
| 332 |
+
# json with segmentations information
|
| 333 |
+
out_file = os.path.join(dataset_dir, f"ADEChallengeData2016/ade20k_panoptic_{name}.json")
|
| 334 |
+
|
| 335 |
+
if not os.path.isdir(out_folder):
|
| 336 |
+
print("Creating folder {} for panoptic segmentation PNGs".format(out_folder))
|
| 337 |
+
os.mkdir(out_folder)
|
| 338 |
+
|
| 339 |
+
# json config
|
| 340 |
+
config_file = "datasets/ade20k_instance_imgCatIds.json"
|
| 341 |
+
with open(config_file) as f:
|
| 342 |
+
config = json.load(f)
|
| 343 |
+
|
| 344 |
+
# load catid mapping
|
| 345 |
+
mapping_file = "datasets/ade20k_instance_catid_mapping.txt"
|
| 346 |
+
with open(mapping_file) as f:
|
| 347 |
+
map_id = {}
|
| 348 |
+
for i, line in enumerate(f.readlines()):
|
| 349 |
+
if i == 0:
|
| 350 |
+
continue
|
| 351 |
+
ins_id, sem_id, _ = line.strip().split()
|
| 352 |
+
# shift id by 1 because we want it to start from 0!
|
| 353 |
+
# ignore_label becomes 255
|
| 354 |
+
map_id[int(ins_id) - 1] = int(sem_id) - 1
|
| 355 |
+
|
| 356 |
+
ADE20K_150_CATEGORIES = []
|
| 357 |
+
for cat_id, cat_name in enumerate(ADE20K_SEM_SEG_CATEGORIES):
|
| 358 |
+
ADE20K_150_CATEGORIES.append(
|
| 359 |
+
{
|
| 360 |
+
"name": cat_name,
|
| 361 |
+
"id": cat_id,
|
| 362 |
+
"isthing": int(cat_id in map_id.values()),
|
| 363 |
+
"color": PALETTE[cat_id],
|
| 364 |
+
}
|
| 365 |
+
)
|
| 366 |
+
categories_dict = {cat["id"]: cat for cat in ADE20K_150_CATEGORIES}
|
| 367 |
+
|
| 368 |
+
panoptic_json_categories = ADE20K_150_CATEGORIES[:]
|
| 369 |
+
panoptic_json_images = []
|
| 370 |
+
panoptic_json_annotations = []
|
| 371 |
+
|
| 372 |
+
filenames = sorted(glob.glob(os.path.join(image_dir, "*.jpg")))
|
| 373 |
+
for idx, filename in enumerate(tqdm.tqdm(filenames)):
|
| 374 |
+
panoptic_json_image = {}
|
| 375 |
+
panoptic_json_annotation = {}
|
| 376 |
+
|
| 377 |
+
image_id = os.path.basename(filename).split(".")[0]
|
| 378 |
+
|
| 379 |
+
panoptic_json_image["id"] = image_id
|
| 380 |
+
panoptic_json_image["file_name"] = os.path.basename(filename)
|
| 381 |
+
|
| 382 |
+
original_format = np.array(Image.open(filename))
|
| 383 |
+
panoptic_json_image["width"] = original_format.shape[1]
|
| 384 |
+
panoptic_json_image["height"] = original_format.shape[0]
|
| 385 |
+
|
| 386 |
+
pan_seg = np.zeros(
|
| 387 |
+
(original_format.shape[0], original_format.shape[1], 3), dtype=np.uint8
|
| 388 |
+
)
|
| 389 |
+
id_generator = IdGenerator(categories_dict)
|
| 390 |
+
|
| 391 |
+
filename_semantic = os.path.join(semantic_dir, image_id + ".png")
|
| 392 |
+
filename_instance = os.path.join(instance_dir, image_id + ".png")
|
| 393 |
+
|
| 394 |
+
sem_seg = np.asarray(Image.open(filename_semantic))
|
| 395 |
+
ins_seg = np.asarray(Image.open(filename_instance))
|
| 396 |
+
|
| 397 |
+
assert sem_seg.dtype == np.uint8
|
| 398 |
+
assert ins_seg.dtype == np.uint8
|
| 399 |
+
|
| 400 |
+
semantic_cat_ids = sem_seg - 1
|
| 401 |
+
instance_cat_ids = ins_seg[..., 0] - 1
|
| 402 |
+
# instance id starts from 1!
|
| 403 |
+
# because 0 is reserved as VOID label
|
| 404 |
+
instance_ins_ids = ins_seg[..., 1]
|
| 405 |
+
|
| 406 |
+
segm_info = []
|
| 407 |
+
|
| 408 |
+
# NOTE: there is some overlap between semantic and instance annotation
|
| 409 |
+
# thus we paste stuffs first
|
| 410 |
+
|
| 411 |
+
# process stuffs
|
| 412 |
+
for semantic_cat_id in np.unique(semantic_cat_ids):
|
| 413 |
+
if semantic_cat_id == 255:
|
| 414 |
+
continue
|
| 415 |
+
if categories_dict[semantic_cat_id]["isthing"]:
|
| 416 |
+
continue
|
| 417 |
+
mask = semantic_cat_ids == semantic_cat_id
|
| 418 |
+
# should not have any overlap
|
| 419 |
+
assert pan_seg[mask].sum() == 0
|
| 420 |
+
|
| 421 |
+
segment_id, color = id_generator.get_id_and_color(semantic_cat_id)
|
| 422 |
+
pan_seg[mask] = color
|
| 423 |
+
|
| 424 |
+
area = np.sum(mask) # segment area computation
|
| 425 |
+
# bbox computation for a segment
|
| 426 |
+
hor = np.sum(mask, axis=0)
|
| 427 |
+
hor_idx = np.nonzero(hor)[0]
|
| 428 |
+
x = hor_idx[0]
|
| 429 |
+
width = hor_idx[-1] - x + 1
|
| 430 |
+
vert = np.sum(mask, axis=1)
|
| 431 |
+
vert_idx = np.nonzero(vert)[0]
|
| 432 |
+
y = vert_idx[0]
|
| 433 |
+
height = vert_idx[-1] - y + 1
|
| 434 |
+
bbox = [int(x), int(y), int(width), int(height)]
|
| 435 |
+
|
| 436 |
+
segm_info.append(
|
| 437 |
+
{
|
| 438 |
+
"id": int(segment_id),
|
| 439 |
+
"category_id": int(semantic_cat_id),
|
| 440 |
+
"area": int(area),
|
| 441 |
+
"bbox": bbox,
|
| 442 |
+
"iscrowd": 0,
|
| 443 |
+
}
|
| 444 |
+
)
|
| 445 |
+
|
| 446 |
+
# process things
|
| 447 |
+
for thing_id in np.unique(instance_ins_ids):
|
| 448 |
+
if thing_id == 0:
|
| 449 |
+
continue
|
| 450 |
+
mask = instance_ins_ids == thing_id
|
| 451 |
+
instance_cat_id = np.unique(instance_cat_ids[mask])
|
| 452 |
+
assert len(instance_cat_id) == 1
|
| 453 |
+
|
| 454 |
+
semantic_cat_id = map_id[instance_cat_id[0]]
|
| 455 |
+
|
| 456 |
+
segment_id, color = id_generator.get_id_and_color(semantic_cat_id)
|
| 457 |
+
pan_seg[mask] = color
|
| 458 |
+
|
| 459 |
+
area = np.sum(mask) # segment area computation
|
| 460 |
+
# bbox computation for a segment
|
| 461 |
+
hor = np.sum(mask, axis=0)
|
| 462 |
+
hor_idx = np.nonzero(hor)[0]
|
| 463 |
+
x = hor_idx[0]
|
| 464 |
+
width = hor_idx[-1] - x + 1
|
| 465 |
+
vert = np.sum(mask, axis=1)
|
| 466 |
+
vert_idx = np.nonzero(vert)[0]
|
| 467 |
+
y = vert_idx[0]
|
| 468 |
+
height = vert_idx[-1] - y + 1
|
| 469 |
+
bbox = [int(x), int(y), int(width), int(height)]
|
| 470 |
+
|
| 471 |
+
segm_info.append(
|
| 472 |
+
{
|
| 473 |
+
"id": int(segment_id),
|
| 474 |
+
"category_id": int(semantic_cat_id),
|
| 475 |
+
"area": int(area),
|
| 476 |
+
"bbox": bbox,
|
| 477 |
+
"iscrowd": 0,
|
| 478 |
+
}
|
| 479 |
+
)
|
| 480 |
+
|
| 481 |
+
panoptic_json_annotation = {
|
| 482 |
+
"image_id": image_id,
|
| 483 |
+
"file_name": image_id + ".png",
|
| 484 |
+
"segments_info": segm_info,
|
| 485 |
+
}
|
| 486 |
+
|
| 487 |
+
Image.fromarray(pan_seg).save(os.path.join(out_folder, image_id + ".png"))
|
| 488 |
+
|
| 489 |
+
panoptic_json_images.append(panoptic_json_image)
|
| 490 |
+
panoptic_json_annotations.append(panoptic_json_annotation)
|
| 491 |
+
|
| 492 |
+
# save this
|
| 493 |
+
d = {
|
| 494 |
+
"images": panoptic_json_images,
|
| 495 |
+
"annotations": panoptic_json_annotations,
|
| 496 |
+
"categories": panoptic_json_categories,
|
| 497 |
+
}
|
| 498 |
+
|
| 499 |
+
save_json(d, out_file)
|
datasets/prepare_ade20k_sem_seg.py
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
+
# -*- coding: utf-8 -*-
|
| 3 |
+
import os
|
| 4 |
+
from pathlib import Path
|
| 5 |
+
|
| 6 |
+
import numpy as np
|
| 7 |
+
import tqdm
|
| 8 |
+
from PIL import Image
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
def convert(input, output):
|
| 12 |
+
img = np.asarray(Image.open(input))
|
| 13 |
+
assert img.dtype == np.uint8
|
| 14 |
+
img = img - 1 # 0 (ignore) becomes 255. others are shifted by 1
|
| 15 |
+
Image.fromarray(img).save(output)
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
if __name__ == "__main__":
|
| 19 |
+
dataset_dir = Path(os.getenv("DETECTRON2_DATASETS", "datasets")) / "ADEChallengeData2016"
|
| 20 |
+
for name in ["training", "validation"]:
|
| 21 |
+
annotation_dir = dataset_dir / "annotations" / name
|
| 22 |
+
output_dir = dataset_dir / "annotations_detectron2" / name
|
| 23 |
+
output_dir.mkdir(parents=True, exist_ok=True)
|
| 24 |
+
for file in tqdm.tqdm(list(annotation_dir.iterdir())):
|
| 25 |
+
output_file = output_dir / file.name
|
| 26 |
+
convert(file, output_file)
|
datasets/prepare_coco_semantic_annos_from_panoptic_annos.py
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env python3
|
| 2 |
+
# -*- coding: utf-8 -*-
|
| 3 |
+
import functools
|
| 4 |
+
import json
|
| 5 |
+
import multiprocessing as mp
|
| 6 |
+
import numpy as np
|
| 7 |
+
import os
|
| 8 |
+
import time
|
| 9 |
+
from fvcore.common.download import download
|
| 10 |
+
from panopticapi.utils import rgb2id
|
| 11 |
+
from PIL import Image
|
| 12 |
+
|
| 13 |
+
from detectron2.data.datasets.builtin_meta import COCO_CATEGORIES
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
def _process_panoptic_to_semantic(input_panoptic, output_semantic, segments, id_map):
|
| 17 |
+
panoptic = np.asarray(Image.open(input_panoptic), dtype=np.uint32)
|
| 18 |
+
panoptic = rgb2id(panoptic)
|
| 19 |
+
output = np.zeros_like(panoptic, dtype=np.uint8) + 255
|
| 20 |
+
for seg in segments:
|
| 21 |
+
cat_id = seg["category_id"]
|
| 22 |
+
new_cat_id = id_map[cat_id]
|
| 23 |
+
output[panoptic == seg["id"]] = new_cat_id
|
| 24 |
+
Image.fromarray(output).save(output_semantic)
|
| 25 |
+
|
| 26 |
+
|
| 27 |
+
def separate_coco_semantic_from_panoptic(panoptic_json, panoptic_root, sem_seg_root, categories):
|
| 28 |
+
"""
|
| 29 |
+
Create semantic segmentation annotations from panoptic segmentation
|
| 30 |
+
annotations, to be used by PanopticFPN.
|
| 31 |
+
It maps all thing categories to class 0, and maps all unlabeled pixels to class 255.
|
| 32 |
+
It maps all stuff categories to contiguous ids starting from 1.
|
| 33 |
+
Args:
|
| 34 |
+
panoptic_json (str): path to the panoptic json file, in COCO's format.
|
| 35 |
+
panoptic_root (str): a directory with panoptic annotation files, in COCO's format.
|
| 36 |
+
sem_seg_root (str): a directory to output semantic annotation files
|
| 37 |
+
categories (list[dict]): category metadata. Each dict needs to have:
|
| 38 |
+
"id": corresponds to the "category_id" in the json annotations
|
| 39 |
+
"isthing": 0 or 1
|
| 40 |
+
"""
|
| 41 |
+
os.makedirs(sem_seg_root, exist_ok=True)
|
| 42 |
+
|
| 43 |
+
id_map = {} # map from category id to id in the output semantic annotation
|
| 44 |
+
assert len(categories) <= 254
|
| 45 |
+
for i, k in enumerate(categories):
|
| 46 |
+
id_map[k["id"]] = i
|
| 47 |
+
# what is id = 0?
|
| 48 |
+
# id_map[0] = 255
|
| 49 |
+
print(id_map)
|
| 50 |
+
|
| 51 |
+
with open(panoptic_json) as f:
|
| 52 |
+
obj = json.load(f)
|
| 53 |
+
|
| 54 |
+
pool = mp.Pool(processes=max(mp.cpu_count() // 2, 4))
|
| 55 |
+
|
| 56 |
+
def iter_annotations():
|
| 57 |
+
for anno in obj["annotations"]:
|
| 58 |
+
file_name = anno["file_name"]
|
| 59 |
+
segments = anno["segments_info"]
|
| 60 |
+
input = os.path.join(panoptic_root, file_name)
|
| 61 |
+
output = os.path.join(sem_seg_root, file_name)
|
| 62 |
+
yield input, output, segments
|
| 63 |
+
|
| 64 |
+
print("Start writing to {} ...".format(sem_seg_root))
|
| 65 |
+
start = time.time()
|
| 66 |
+
pool.starmap(
|
| 67 |
+
functools.partial(_process_panoptic_to_semantic, id_map=id_map),
|
| 68 |
+
iter_annotations(),
|
| 69 |
+
chunksize=100,
|
| 70 |
+
)
|
| 71 |
+
print("Finished. time: {:.2f}s".format(time.time() - start))
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
if __name__ == "__main__":
|
| 75 |
+
dataset_dir = os.path.join(os.getenv("DETECTRON2_DATASETS", "datasets"), "coco")
|
| 76 |
+
for s in ["val2017", "train2017"]:
|
| 77 |
+
separate_coco_semantic_from_panoptic(
|
| 78 |
+
os.path.join(dataset_dir, "annotations/panoptic_{}.json".format(s)),
|
| 79 |
+
os.path.join(dataset_dir, "panoptic_{}".format(s)),
|
| 80 |
+
os.path.join(dataset_dir, "panoptic_semseg_{}".format(s)),
|
| 81 |
+
COCO_CATEGORIES,
|
| 82 |
+
)
|
datasets/prepare_pascal_ctx_full_sem_seg.py
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import numpy as np
|
| 3 |
+
from pathlib import Path
|
| 4 |
+
from PIL import Image
|
| 5 |
+
import scipy.io as sio
|
| 6 |
+
|
| 7 |
+
import tqdm
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
def generate_labels(mat_file, out_dir):
|
| 11 |
+
|
| 12 |
+
mat = sio.loadmat(mat_file)
|
| 13 |
+
label_map = mat["LabelMap"]
|
| 14 |
+
assert label_map.dtype == np.uint16
|
| 15 |
+
label_map[label_map == 0] = 65535
|
| 16 |
+
label_map = label_map - 1
|
| 17 |
+
label_map[label_map == 65534] = 65535
|
| 18 |
+
|
| 19 |
+
out_file = out_dir / Path(mat_file.name).with_suffix(".tif")
|
| 20 |
+
Image.fromarray(label_map).save(out_file)
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
if __name__ == "__main__":
|
| 24 |
+
dataset_dir = Path(os.getenv("DETECTRON2_DATASETS", "datasets")) / "pascal_ctx_d2"
|
| 25 |
+
voc_dir = Path(os.getenv("DETECTRON2_DATASETS", "datasets")) / "VOCdevkit/VOC2010"
|
| 26 |
+
mat_dir = voc_dir / "trainval"
|
| 27 |
+
for split in ["training", "validation"]:
|
| 28 |
+
file_names = list((dataset_dir / "images" / split).glob("*.jpg"))
|
| 29 |
+
output_img_dir = dataset_dir / "images" / split
|
| 30 |
+
output_ann_dir = dataset_dir / "annotations_ctx459" / split
|
| 31 |
+
|
| 32 |
+
output_img_dir.mkdir(parents=True, exist_ok=True)
|
| 33 |
+
output_ann_dir.mkdir(parents=True, exist_ok=True)
|
| 34 |
+
|
| 35 |
+
for file_name in tqdm.tqdm(file_names):
|
| 36 |
+
mat_file_path = mat_dir / f"{file_name.stem}.mat"
|
| 37 |
+
|
| 38 |
+
generate_labels(mat_file_path, output_ann_dir)
|
datasets/prepare_pascal_ctx_sem_seg.py
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
from pathlib import Path
|
| 3 |
+
import shutil
|
| 4 |
+
|
| 5 |
+
import numpy as np
|
| 6 |
+
import tqdm
|
| 7 |
+
from PIL import Image
|
| 8 |
+
import multiprocessing as mp
|
| 9 |
+
import functools
|
| 10 |
+
from detail import Detail
|
| 11 |
+
|
| 12 |
+
# fmt: off
|
| 13 |
+
_mapping = np.sort(
|
| 14 |
+
np.array([
|
| 15 |
+
0, 2, 259, 260, 415, 324, 9, 258, 144, 18, 19, 22, 23, 397, 25, 284,
|
| 16 |
+
158, 159, 416, 33, 162, 420, 454, 295, 296, 427, 44, 45, 46, 308, 59,
|
| 17 |
+
440, 445, 31, 232, 65, 354, 424, 68, 326, 72, 458, 34, 207, 80, 355,
|
| 18 |
+
85, 347, 220, 349, 360, 98, 187, 104, 105, 366, 189, 368, 113, 115
|
| 19 |
+
]))
|
| 20 |
+
# fmt: on
|
| 21 |
+
_key = np.array(range(len(_mapping))).astype("uint8")
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
def generate_labels(img_info, detail_api, out_dir):
|
| 25 |
+
def _class_to_index(mask, _mapping, _key):
|
| 26 |
+
# assert the values
|
| 27 |
+
values = np.unique(mask)
|
| 28 |
+
for i in range(len(values)):
|
| 29 |
+
assert values[i] in _mapping
|
| 30 |
+
index = np.digitize(mask.ravel(), _mapping, right=True)
|
| 31 |
+
return _key[index].reshape(mask.shape)
|
| 32 |
+
|
| 33 |
+
sem_seg = _class_to_index(detail_api.getMask(img_info), _mapping=_mapping, _key=_key)
|
| 34 |
+
sem_seg = sem_seg - 1 # 0 (ignore) becomes 255. others are shifted by 1
|
| 35 |
+
filename = img_info["file_name"]
|
| 36 |
+
|
| 37 |
+
Image.fromarray(sem_seg).save(out_dir / filename.replace("jpg", "png"))
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
def copy_images(img_info, img_dir, out_dir):
|
| 41 |
+
filename = img_info["file_name"]
|
| 42 |
+
shutil.copy2(img_dir / filename, out_dir / filename)
|
| 43 |
+
|
| 44 |
+
|
| 45 |
+
if __name__ == "__main__":
|
| 46 |
+
dataset_dir = Path(os.getenv("DETECTRON2_DATASETS", "datasets")) / "pascal_ctx_d2"
|
| 47 |
+
voc_dir = Path(os.getenv("DETECTRON2_DATASETS", "datasets")) / "VOCdevkit/VOC2010"
|
| 48 |
+
for split in ["training", "validation"]:
|
| 49 |
+
img_dir = voc_dir / "JPEGImages"
|
| 50 |
+
if split == "training":
|
| 51 |
+
detail_api = Detail(voc_dir / "trainval_merged.json", img_dir, "train")
|
| 52 |
+
else:
|
| 53 |
+
detail_api = Detail(voc_dir / "trainval_merged.json", img_dir, "val")
|
| 54 |
+
img_infos = detail_api.getImgs()
|
| 55 |
+
|
| 56 |
+
output_img_dir = dataset_dir / "images" / split
|
| 57 |
+
output_ann_dir = dataset_dir / "annotations_ctx59" / split
|
| 58 |
+
|
| 59 |
+
output_img_dir.mkdir(parents=True, exist_ok=True)
|
| 60 |
+
output_ann_dir.mkdir(parents=True, exist_ok=True)
|
| 61 |
+
|
| 62 |
+
pool = mp.Pool(processes=max(mp.cpu_count() // 2, 4))
|
| 63 |
+
|
| 64 |
+
pool.map(
|
| 65 |
+
functools.partial(copy_images, img_dir=img_dir, out_dir=output_img_dir),
|
| 66 |
+
tqdm.tqdm(img_infos, desc=f"Writing {split} images to {output_img_dir} ..."),
|
| 67 |
+
chunksize=100,
|
| 68 |
+
)
|
| 69 |
+
|
| 70 |
+
pool.map(
|
| 71 |
+
functools.partial(generate_labels, detail_api=detail_api, out_dir=output_ann_dir),
|
| 72 |
+
tqdm.tqdm(img_infos, desc=f"Writing {split} images to {output_ann_dir} ..."),
|
| 73 |
+
chunksize=100,
|
| 74 |
+
)
|
datasets/prepare_pascal_voc_sem_seg.py
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
from pathlib import Path
|
| 3 |
+
import shutil
|
| 4 |
+
|
| 5 |
+
import numpy as np
|
| 6 |
+
import tqdm
|
| 7 |
+
from PIL import Image
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
def convert_pas21(input, output):
|
| 11 |
+
img = np.asarray(Image.open(input))
|
| 12 |
+
assert img.dtype == np.uint8
|
| 13 |
+
# do nothing
|
| 14 |
+
Image.fromarray(img).save(output)
|
| 15 |
+
|
| 16 |
+
def convert_pas20(input, output):
|
| 17 |
+
img = np.array(Image.open(input))
|
| 18 |
+
img[img == 0] = 255
|
| 19 |
+
img = img - 1
|
| 20 |
+
img[img == 254] = 255
|
| 21 |
+
assert img.dtype == np.uint8
|
| 22 |
+
# do nothing
|
| 23 |
+
Image.fromarray(img).save(output)
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
if __name__ == "__main__":
|
| 27 |
+
dataset_dir = Path(os.getenv("DETECTRON2_DATASETS", "datasets")) / "pascal_voc_d2"
|
| 28 |
+
voc_dir = Path(os.getenv("DETECTRON2_DATASETS", "datasets")) / "VOCdevkit/VOC2012"
|
| 29 |
+
for split in ["training", "validation"]:
|
| 30 |
+
if split == "training":
|
| 31 |
+
img_name_path = voc_dir / "ImageSets/Segmentation/train.txt"
|
| 32 |
+
else:
|
| 33 |
+
img_name_path = voc_dir / "ImageSets/Segmentation/val.txt"
|
| 34 |
+
img_dir = voc_dir / "JPEGImages"
|
| 35 |
+
ann_dir = voc_dir / "SegmentationClass"
|
| 36 |
+
|
| 37 |
+
output_img_dir = dataset_dir / "images" / split
|
| 38 |
+
output_ann_dir_21 = dataset_dir / "annotations_pascal21" / split
|
| 39 |
+
output_ann_dir_20 = dataset_dir / "annotations_pascal20" / split
|
| 40 |
+
|
| 41 |
+
output_img_dir.mkdir(parents=True, exist_ok=True)
|
| 42 |
+
output_ann_dir_21.mkdir(parents=True, exist_ok=True)
|
| 43 |
+
output_ann_dir_20.mkdir(parents=True, exist_ok=True)
|
| 44 |
+
|
| 45 |
+
with open(img_name_path) as f:
|
| 46 |
+
for line in tqdm.tqdm(f.readlines()):
|
| 47 |
+
img_name = line.strip()
|
| 48 |
+
img_path = img_dir / f"{img_name}.jpg"
|
| 49 |
+
ann_path = ann_dir / f"{img_name}.png"
|
| 50 |
+
|
| 51 |
+
# print(f'copy2 {output_img_dir}')
|
| 52 |
+
shutil.copy2(img_path, output_img_dir)
|
| 53 |
+
# print(f"convert {ann_dir} to {output_ann_dir / f'{img_name}.png'}")
|
| 54 |
+
convert_pas21(ann_path, output_ann_dir_21 / f"{img_name}.png")
|
| 55 |
+
convert_pas20(ann_path, output_ann_dir_20 / f"{img_name}.png")
|
demo/demo.py
ADDED
|
@@ -0,0 +1,189 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import argparse
|
| 2 |
+
import glob
|
| 3 |
+
import multiprocessing as mp
|
| 4 |
+
import os
|
| 5 |
+
|
| 6 |
+
# fmt: off
|
| 7 |
+
import sys
|
| 8 |
+
sys.path.insert(1, os.path.join(sys.path[0], '..'))
|
| 9 |
+
# fmt: on
|
| 10 |
+
|
| 11 |
+
import tempfile
|
| 12 |
+
import time
|
| 13 |
+
import warnings
|
| 14 |
+
|
| 15 |
+
import cv2
|
| 16 |
+
import numpy as np
|
| 17 |
+
import tqdm
|
| 18 |
+
|
| 19 |
+
from detectron2.config import get_cfg
|
| 20 |
+
from detectron2.data.detection_utils import read_image
|
| 21 |
+
from detectron2.utils.logger import setup_logger
|
| 22 |
+
|
| 23 |
+
from frozenseg import add_maskformer2_config, add_frozenseg_config
|
| 24 |
+
from predictor import VisualizationDemo
|
| 25 |
+
|
| 26 |
+
# constants
|
| 27 |
+
WINDOW_NAME = "frozenseg demo"
|
| 28 |
+
|
| 29 |
+
|
| 30 |
+
def setup_cfg(args):
|
| 31 |
+
# load config from file and command-line arguments
|
| 32 |
+
cfg = get_cfg()
|
| 33 |
+
add_maskformer2_config(cfg)
|
| 34 |
+
add_frozenseg_config(cfg)
|
| 35 |
+
cfg.merge_from_file(args.config_file)
|
| 36 |
+
cfg.merge_from_list(args.opts)
|
| 37 |
+
cfg.freeze()
|
| 38 |
+
return cfg
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
def get_parser():
|
| 42 |
+
parser = argparse.ArgumentParser(description="frozenseg demo for builtin configs")
|
| 43 |
+
parser.add_argument(
|
| 44 |
+
"--config-file",
|
| 45 |
+
default="configs/coco/frozenseg/convnext_large_eval_ade20k.yaml",
|
| 46 |
+
metavar="FILE",
|
| 47 |
+
help="path to config file",
|
| 48 |
+
)
|
| 49 |
+
parser.add_argument("--webcam", action="store_true", help="Take inputs from webcam.")
|
| 50 |
+
parser.add_argument("--video-input", help="Path to video file.")
|
| 51 |
+
parser.add_argument(
|
| 52 |
+
"--input",
|
| 53 |
+
nargs="+",
|
| 54 |
+
help="A list of space separated input images; "
|
| 55 |
+
"or a single glob pattern such as 'directory/*.jpg'",
|
| 56 |
+
)
|
| 57 |
+
parser.add_argument(
|
| 58 |
+
"--output",
|
| 59 |
+
help="A file or directory to save output visualizations. "
|
| 60 |
+
"If not given, will show output in an OpenCV window.",
|
| 61 |
+
)
|
| 62 |
+
|
| 63 |
+
parser.add_argument(
|
| 64 |
+
"--confidence-threshold",
|
| 65 |
+
type=float,
|
| 66 |
+
default=0.5,
|
| 67 |
+
help="Minimum score for instance predictions to be shown",
|
| 68 |
+
)
|
| 69 |
+
parser.add_argument(
|
| 70 |
+
"--opts",
|
| 71 |
+
help="Modify config options using the command-line 'KEY VALUE' pairs",
|
| 72 |
+
default=[],
|
| 73 |
+
nargs=argparse.REMAINDER,
|
| 74 |
+
)
|
| 75 |
+
return parser
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
def test_opencv_video_format(codec, file_ext):
|
| 79 |
+
with tempfile.TemporaryDirectory(prefix="video_format_test") as dir:
|
| 80 |
+
filename = os.path.join(dir, "test_file" + file_ext)
|
| 81 |
+
writer = cv2.VideoWriter(
|
| 82 |
+
filename=filename,
|
| 83 |
+
fourcc=cv2.VideoWriter_fourcc(*codec),
|
| 84 |
+
fps=float(30),
|
| 85 |
+
frameSize=(10, 10),
|
| 86 |
+
isColor=True,
|
| 87 |
+
)
|
| 88 |
+
[writer.write(np.zeros((10, 10, 3), np.uint8)) for _ in range(30)]
|
| 89 |
+
writer.release()
|
| 90 |
+
if os.path.isfile(filename):
|
| 91 |
+
return True
|
| 92 |
+
return False
|
| 93 |
+
|
| 94 |
+
|
| 95 |
+
if __name__ == "__main__":
|
| 96 |
+
mp.set_start_method("spawn", force=True)
|
| 97 |
+
args = get_parser().parse_args()
|
| 98 |
+
setup_logger(name="fvcore")
|
| 99 |
+
logger = setup_logger()
|
| 100 |
+
logger.info("Arguments: " + str(args))
|
| 101 |
+
|
| 102 |
+
cfg = setup_cfg(args)
|
| 103 |
+
demo = VisualizationDemo(cfg)
|
| 104 |
+
|
| 105 |
+
if args.input:
|
| 106 |
+
if len(args.input) == 1:
|
| 107 |
+
args.input = glob.glob(os.path.expanduser(args.input[0]))
|
| 108 |
+
assert args.input, "The input path(s) was not found"
|
| 109 |
+
for path in tqdm.tqdm(args.input, disable=not args.output):
|
| 110 |
+
# use PIL, to be consistent with evaluation
|
| 111 |
+
img = read_image(path, format="BGR")
|
| 112 |
+
start_time = time.time()
|
| 113 |
+
predictions, visualized_output = demo.run_on_image(img)
|
| 114 |
+
logger.info(
|
| 115 |
+
"{}: {} in {:.2f}s".format(
|
| 116 |
+
path,
|
| 117 |
+
"detected {} instances".format(len(predictions["instances"]))
|
| 118 |
+
if "instances" in predictions
|
| 119 |
+
else "finished",
|
| 120 |
+
time.time() - start_time,
|
| 121 |
+
)
|
| 122 |
+
)
|
| 123 |
+
|
| 124 |
+
if args.output:
|
| 125 |
+
if os.path.isdir(args.output):
|
| 126 |
+
assert os.path.isdir(args.output), args.output
|
| 127 |
+
out_filename = os.path.join(args.output, os.path.basename(path))
|
| 128 |
+
else:
|
| 129 |
+
assert len(args.input) == 1, "Please specify a directory with args.output"
|
| 130 |
+
out_filename = args.output
|
| 131 |
+
visualized_output.save(out_filename)
|
| 132 |
+
else:
|
| 133 |
+
cv2.namedWindow(WINDOW_NAME, cv2.WINDOW_NORMAL)
|
| 134 |
+
cv2.imshow(WINDOW_NAME, visualized_output.get_image()[:, :, ::-1])
|
| 135 |
+
if cv2.waitKey(0) == 27:
|
| 136 |
+
break # esc to quit
|
| 137 |
+
elif args.webcam:
|
| 138 |
+
assert args.input is None, "Cannot have both --input and --webcam!"
|
| 139 |
+
assert args.output is None, "output not yet supported with --webcam!"
|
| 140 |
+
cam = cv2.VideoCapture(0)
|
| 141 |
+
for vis in tqdm.tqdm(demo.run_on_video(cam)):
|
| 142 |
+
cv2.namedWindow(WINDOW_NAME, cv2.WINDOW_NORMAL)
|
| 143 |
+
cv2.imshow(WINDOW_NAME, vis)
|
| 144 |
+
if cv2.waitKey(1) == 27:
|
| 145 |
+
break # esc to quit
|
| 146 |
+
cam.release()
|
| 147 |
+
cv2.destroyAllWindows()
|
| 148 |
+
elif args.video_input:
|
| 149 |
+
video = cv2.VideoCapture(args.video_input)
|
| 150 |
+
width = int(video.get(cv2.CAP_PROP_FRAME_WIDTH))
|
| 151 |
+
height = int(video.get(cv2.CAP_PROP_FRAME_HEIGHT))
|
| 152 |
+
frames_per_second = video.get(cv2.CAP_PROP_FPS)
|
| 153 |
+
num_frames = int(video.get(cv2.CAP_PROP_FRAME_COUNT))
|
| 154 |
+
basename = os.path.basename(args.video_input)
|
| 155 |
+
codec, file_ext = (
|
| 156 |
+
("x264", ".mkv") if test_opencv_video_format("x264", ".mkv") else ("mp4v", ".mp4")
|
| 157 |
+
)
|
| 158 |
+
if codec == ".mp4v":
|
| 159 |
+
warnings.warn("x264 codec not available, switching to mp4v")
|
| 160 |
+
if args.output:
|
| 161 |
+
if os.path.isdir(args.output):
|
| 162 |
+
output_fname = os.path.join(args.output, basename)
|
| 163 |
+
output_fname = os.path.splitext(output_fname)[0] + file_ext
|
| 164 |
+
else:
|
| 165 |
+
output_fname = args.output
|
| 166 |
+
assert not os.path.isfile(output_fname), output_fname
|
| 167 |
+
output_file = cv2.VideoWriter(
|
| 168 |
+
filename=output_fname,
|
| 169 |
+
# some installation of opencv may not support x264 (due to its license),
|
| 170 |
+
# you can try other format (e.g. MPEG)
|
| 171 |
+
fourcc=cv2.VideoWriter_fourcc(*codec),
|
| 172 |
+
fps=float(frames_per_second),
|
| 173 |
+
frameSize=(width, height),
|
| 174 |
+
isColor=True,
|
| 175 |
+
)
|
| 176 |
+
assert os.path.isfile(args.video_input)
|
| 177 |
+
for vis_frame in tqdm.tqdm(demo.run_on_video(video), total=num_frames):
|
| 178 |
+
if args.output:
|
| 179 |
+
output_file.write(vis_frame)
|
| 180 |
+
else:
|
| 181 |
+
cv2.namedWindow(basename, cv2.WINDOW_NORMAL)
|
| 182 |
+
cv2.imshow(basename, vis_frame)
|
| 183 |
+
if cv2.waitKey(1) == 27:
|
| 184 |
+
break # esc to quit
|
| 185 |
+
video.release()
|
| 186 |
+
if args.output:
|
| 187 |
+
output_file.release()
|
| 188 |
+
else:
|
| 189 |
+
cv2.destroyAllWindows()
|
demo/predictor.py
ADDED
|
@@ -0,0 +1,273 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import atexit
|
| 2 |
+
import bisect
|
| 3 |
+
import multiprocessing as mp
|
| 4 |
+
from collections import deque
|
| 5 |
+
|
| 6 |
+
import cv2
|
| 7 |
+
import torch
|
| 8 |
+
import itertools
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
from detectron2.data import DatasetCatalog, MetadataCatalog
|
| 12 |
+
from detectron2.engine.defaults import DefaultPredictor as d2_defaultPredictor
|
| 13 |
+
from detectron2.utils.video_visualizer import VideoVisualizer
|
| 14 |
+
from detectron2.utils.visualizer import ColorMode, Visualizer, random_color
|
| 15 |
+
import detectron2.utils.visualizer as d2_visualizer
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
class DefaultPredictor(d2_defaultPredictor):
|
| 19 |
+
|
| 20 |
+
def set_metadata(self, metadata):
|
| 21 |
+
self.model.set_metadata(metadata)
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
class OpenVocabVisualizer(Visualizer):
|
| 25 |
+
def draw_panoptic_seg(self, panoptic_seg, segments_info, area_threshold=None, alpha=0.7):
|
| 26 |
+
"""
|
| 27 |
+
Draw panoptic prediction annotations or results.
|
| 28 |
+
|
| 29 |
+
Args:
|
| 30 |
+
panoptic_seg (Tensor): of shape (height, width) where the values are ids for each
|
| 31 |
+
segment.
|
| 32 |
+
segments_info (list[dict] or None): Describe each segment in `panoptic_seg`.
|
| 33 |
+
If it is a ``list[dict]``, each dict contains keys "id", "category_id".
|
| 34 |
+
If None, category id of each pixel is computed by
|
| 35 |
+
``pixel // metadata.label_divisor``.
|
| 36 |
+
area_threshold (int): stuff segments with less than `area_threshold` are not drawn.
|
| 37 |
+
|
| 38 |
+
Returns:
|
| 39 |
+
output (VisImage): image object with visualizations.
|
| 40 |
+
"""
|
| 41 |
+
pred = d2_visualizer._PanopticPrediction(panoptic_seg, segments_info, self.metadata)
|
| 42 |
+
|
| 43 |
+
if self._instance_mode == ColorMode.IMAGE_BW:
|
| 44 |
+
self.output.reset_image(self._create_grayscale_image(pred.non_empty_mask()))
|
| 45 |
+
# draw mask for all semantic segments first i.e. "stuff"
|
| 46 |
+
for mask, sinfo in pred.semantic_masks():
|
| 47 |
+
category_idx = sinfo["category_id"]
|
| 48 |
+
try:
|
| 49 |
+
mask_color = [x / 255 for x in self.metadata.stuff_colors[category_idx]]
|
| 50 |
+
except AttributeError:
|
| 51 |
+
mask_color = None
|
| 52 |
+
|
| 53 |
+
text = self.metadata.stuff_classes[category_idx].split(',')[0]
|
| 54 |
+
self.draw_binary_mask(
|
| 55 |
+
mask,
|
| 56 |
+
color=mask_color,
|
| 57 |
+
edge_color=d2_visualizer._OFF_WHITE,
|
| 58 |
+
text=text,
|
| 59 |
+
alpha=alpha,
|
| 60 |
+
area_threshold=area_threshold,
|
| 61 |
+
)
|
| 62 |
+
# draw mask for all instances second
|
| 63 |
+
all_instances = list(pred.instance_masks())
|
| 64 |
+
if len(all_instances) == 0:
|
| 65 |
+
return self.output
|
| 66 |
+
masks, sinfo = list(zip(*all_instances))
|
| 67 |
+
category_ids = [x["category_id"] for x in sinfo]
|
| 68 |
+
|
| 69 |
+
try:
|
| 70 |
+
scores = [x["score"] for x in sinfo]
|
| 71 |
+
except KeyError:
|
| 72 |
+
scores = None
|
| 73 |
+
stuff_classes = self.metadata.stuff_classes
|
| 74 |
+
stuff_classes = [x.split(',')[0] for x in stuff_classes]
|
| 75 |
+
labels = d2_visualizer._create_text_labels(
|
| 76 |
+
category_ids, scores, stuff_classes, [x.get("iscrowd", 0) for x in sinfo]
|
| 77 |
+
)
|
| 78 |
+
|
| 79 |
+
try:
|
| 80 |
+
colors = [
|
| 81 |
+
self._jitter([x / 255 for x in self.metadata.stuff_colors[c]]) for c in category_ids
|
| 82 |
+
]
|
| 83 |
+
except AttributeError:
|
| 84 |
+
colors = None
|
| 85 |
+
self.overlay_instances(masks=masks, labels=labels, assigned_colors=colors, alpha=alpha)
|
| 86 |
+
|
| 87 |
+
return self.output
|
| 88 |
+
|
| 89 |
+
|
| 90 |
+
class VisualizationDemo(object):
|
| 91 |
+
def __init__(self, cfg, instance_mode=ColorMode.IMAGE, parallel=False):
|
| 92 |
+
"""
|
| 93 |
+
Args:
|
| 94 |
+
cfg (CfgNode):
|
| 95 |
+
instance_mode (ColorMode):
|
| 96 |
+
parallel (bool): whether to run the model in different processes from visualization.
|
| 97 |
+
Useful since the visualization logic can be slow.
|
| 98 |
+
"""
|
| 99 |
+
|
| 100 |
+
coco_metadata = MetadataCatalog.get("openvocab_coco_2017_val_panoptic_with_sem_seg")
|
| 101 |
+
ade20k_metadata = MetadataCatalog.get("openvocab_ade20k_panoptic_val")
|
| 102 |
+
lvis_classes = open("./frozenseg/data/datasets/lvis_1203_with_prompt_eng.txt", 'r').read().splitlines()
|
| 103 |
+
lvis_classes = [x[x.find(':')+1:] for x in lvis_classes]
|
| 104 |
+
lvis_colors = list(
|
| 105 |
+
itertools.islice(itertools.cycle(coco_metadata.stuff_colors), len(lvis_classes))
|
| 106 |
+
)
|
| 107 |
+
# rerrange to thing_classes, stuff_classes
|
| 108 |
+
coco_thing_classes = coco_metadata.thing_classes
|
| 109 |
+
coco_stuff_classes = [x for x in coco_metadata.stuff_classes if x not in coco_thing_classes]
|
| 110 |
+
coco_thing_colors = coco_metadata.thing_colors
|
| 111 |
+
coco_stuff_colors = [x for x in coco_metadata.stuff_colors if x not in coco_thing_colors]
|
| 112 |
+
ade20k_thing_classes = ade20k_metadata.thing_classes
|
| 113 |
+
ade20k_stuff_classes = [x for x in ade20k_metadata.stuff_classes if x not in ade20k_thing_classes]
|
| 114 |
+
ade20k_thing_colors = ade20k_metadata.thing_colors
|
| 115 |
+
ade20k_stuff_colors = [x for x in ade20k_metadata.stuff_colors if x not in ade20k_thing_colors]
|
| 116 |
+
|
| 117 |
+
user_classes = []
|
| 118 |
+
user_colors = [random_color(rgb=True, maximum=1) for _ in range(len(user_classes))]
|
| 119 |
+
|
| 120 |
+
stuff_classes = coco_stuff_classes + ade20k_stuff_classes
|
| 121 |
+
stuff_colors = coco_stuff_colors + ade20k_stuff_colors
|
| 122 |
+
thing_classes = user_classes + coco_thing_classes + ade20k_thing_classes + lvis_classes
|
| 123 |
+
thing_colors = user_colors + coco_thing_colors + ade20k_thing_colors + lvis_colors
|
| 124 |
+
|
| 125 |
+
thing_dataset_id_to_contiguous_id = {x: x for x in range(len(thing_classes))}
|
| 126 |
+
DatasetCatalog.register(
|
| 127 |
+
"openvocab_dataset", lambda x: []
|
| 128 |
+
)
|
| 129 |
+
self.metadata = MetadataCatalog.get("openvocab_dataset").set(
|
| 130 |
+
stuff_classes=thing_classes+stuff_classes,
|
| 131 |
+
stuff_colors=thing_colors+stuff_colors,
|
| 132 |
+
thing_dataset_id_to_contiguous_id=thing_dataset_id_to_contiguous_id,
|
| 133 |
+
)
|
| 134 |
+
#print("self.metadata:", self.metadata)
|
| 135 |
+
self.cpu_device = torch.device("cpu")
|
| 136 |
+
self.instance_mode = instance_mode
|
| 137 |
+
|
| 138 |
+
self.parallel = parallel
|
| 139 |
+
if parallel:
|
| 140 |
+
num_gpu = torch.cuda.device_count()
|
| 141 |
+
self.predictor = AsyncPredictor(cfg, num_gpus=num_gpu)
|
| 142 |
+
else:
|
| 143 |
+
self.predictor = DefaultPredictor(cfg)
|
| 144 |
+
self.predictor.set_metadata(self.metadata)
|
| 145 |
+
|
| 146 |
+
def run_on_image(self, image):
|
| 147 |
+
"""
|
| 148 |
+
Args:
|
| 149 |
+
image (np.ndarray): an image of shape (H, W, C) (in BGR order).
|
| 150 |
+
This is the format used by OpenCV.
|
| 151 |
+
Returns:
|
| 152 |
+
predictions (dict): the output of the model.
|
| 153 |
+
vis_output (VisImage): the visualized image output.
|
| 154 |
+
"""
|
| 155 |
+
vis_output = None
|
| 156 |
+
predictions = self.predictor(image)
|
| 157 |
+
# Convert image from OpenCV BGR format to Matplotlib RGB format.
|
| 158 |
+
image = image[:, :, ::-1]
|
| 159 |
+
visualizer = OpenVocabVisualizer(image, self.metadata, instance_mode=self.instance_mode)
|
| 160 |
+
if "panoptic_seg" in predictions:
|
| 161 |
+
panoptic_seg, segments_info = predictions["panoptic_seg"]
|
| 162 |
+
vis_output = visualizer.draw_panoptic_seg(
|
| 163 |
+
panoptic_seg.to(self.cpu_device), segments_info
|
| 164 |
+
)
|
| 165 |
+
else:
|
| 166 |
+
if "sem_seg" in predictions:
|
| 167 |
+
vis_output = visualizer.draw_sem_seg(
|
| 168 |
+
predictions["sem_seg"].argmax(dim=0).to(self.cpu_device)
|
| 169 |
+
)
|
| 170 |
+
if "instances" in predictions:
|
| 171 |
+
instances = predictions["instances"].to(self.cpu_device)
|
| 172 |
+
vis_output = visualizer.draw_instance_predictions(predictions=instances)
|
| 173 |
+
|
| 174 |
+
return predictions, vis_output
|
| 175 |
+
|
| 176 |
+
def _frame_from_video(self, video):
|
| 177 |
+
while video.isOpened():
|
| 178 |
+
success, frame = video.read()
|
| 179 |
+
if success:
|
| 180 |
+
yield frame
|
| 181 |
+
else:
|
| 182 |
+
break
|
| 183 |
+
|
| 184 |
+
|
| 185 |
+
class AsyncPredictor:
|
| 186 |
+
"""
|
| 187 |
+
A predictor that runs the model asynchronously, possibly on >1 GPUs.
|
| 188 |
+
Because rendering the visualization takes considerably amount of time,
|
| 189 |
+
this helps improve throughput a little bit when rendering videos.
|
| 190 |
+
"""
|
| 191 |
+
|
| 192 |
+
class _StopToken:
|
| 193 |
+
pass
|
| 194 |
+
|
| 195 |
+
class _PredictWorker(mp.Process):
|
| 196 |
+
def __init__(self, cfg, task_queue, result_queue):
|
| 197 |
+
self.cfg = cfg
|
| 198 |
+
self.task_queue = task_queue
|
| 199 |
+
self.result_queue = result_queue
|
| 200 |
+
super().__init__()
|
| 201 |
+
|
| 202 |
+
def run(self):
|
| 203 |
+
predictor = DefaultPredictor(self.cfg)
|
| 204 |
+
|
| 205 |
+
while True:
|
| 206 |
+
task = self.task_queue.get()
|
| 207 |
+
if isinstance(task, AsyncPredictor._StopToken):
|
| 208 |
+
break
|
| 209 |
+
idx, data = task
|
| 210 |
+
result = predictor(data)
|
| 211 |
+
self.result_queue.put((idx, result))
|
| 212 |
+
|
| 213 |
+
def __init__(self, cfg, num_gpus: int = 1):
|
| 214 |
+
"""
|
| 215 |
+
Args:
|
| 216 |
+
cfg (CfgNode):
|
| 217 |
+
num_gpus (int): if 0, will run on CPU
|
| 218 |
+
"""
|
| 219 |
+
num_workers = max(num_gpus, 1)
|
| 220 |
+
self.task_queue = mp.Queue(maxsize=num_workers * 3)
|
| 221 |
+
self.result_queue = mp.Queue(maxsize=num_workers * 3)
|
| 222 |
+
self.procs = []
|
| 223 |
+
for gpuid in range(max(num_gpus, 1)):
|
| 224 |
+
cfg = cfg.clone()
|
| 225 |
+
cfg.defrost()
|
| 226 |
+
cfg.MODEL.DEVICE = "cuda:{}".format(gpuid) if num_gpus > 0 else "cpu"
|
| 227 |
+
self.procs.append(
|
| 228 |
+
AsyncPredictor._PredictWorker(cfg, self.task_queue, self.result_queue)
|
| 229 |
+
)
|
| 230 |
+
|
| 231 |
+
self.put_idx = 0
|
| 232 |
+
self.get_idx = 0
|
| 233 |
+
self.result_rank = []
|
| 234 |
+
self.result_data = []
|
| 235 |
+
|
| 236 |
+
for p in self.procs:
|
| 237 |
+
p.start()
|
| 238 |
+
atexit.register(self.shutdown)
|
| 239 |
+
|
| 240 |
+
def put(self, image):
|
| 241 |
+
self.put_idx += 1
|
| 242 |
+
self.task_queue.put((self.put_idx, image))
|
| 243 |
+
|
| 244 |
+
def get(self):
|
| 245 |
+
self.get_idx += 1 # the index needed for this request
|
| 246 |
+
if len(self.result_rank) and self.result_rank[0] == self.get_idx:
|
| 247 |
+
res = self.result_data[0]
|
| 248 |
+
del self.result_data[0], self.result_rank[0]
|
| 249 |
+
return res
|
| 250 |
+
|
| 251 |
+
while True:
|
| 252 |
+
# make sure the results are returned in the correct order
|
| 253 |
+
idx, res = self.result_queue.get()
|
| 254 |
+
if idx == self.get_idx:
|
| 255 |
+
return res
|
| 256 |
+
insert = bisect.bisect(self.result_rank, idx)
|
| 257 |
+
self.result_rank.insert(insert, idx)
|
| 258 |
+
self.result_data.insert(insert, res)
|
| 259 |
+
|
| 260 |
+
def __len__(self):
|
| 261 |
+
return self.put_idx - self.get_idx
|
| 262 |
+
|
| 263 |
+
def __call__(self, image):
|
| 264 |
+
self.put(image)
|
| 265 |
+
return self.get()
|
| 266 |
+
|
| 267 |
+
def shutdown(self):
|
| 268 |
+
for _ in self.procs:
|
| 269 |
+
self.task_queue.put(AsyncPredictor._StopToken())
|
| 270 |
+
|
| 271 |
+
@property
|
| 272 |
+
def default_buffer_size(self):
|
| 273 |
+
return len(self.procs) * 5
|
eval.sh
ADDED
|
@@ -0,0 +1,70 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/bin/bash
|
| 2 |
+
#SBATCH --job-name=frozenseg_eval
|
| 3 |
+
#SBATCH --output=output/slurm/%j.run.out
|
| 4 |
+
#SBATCH --error=output/slurm/%j.run.err
|
| 5 |
+
#SBATCH --partition=gpu-a100
|
| 6 |
+
#SBATCH --gres=gpu:1
|
| 7 |
+
#SBATCH --cpus-per-task=16
|
| 8 |
+
#SBATCH --comment=yhx_team
|
| 9 |
+
export MODULEPATH="/opt/app/spack/share/spack/modules/linux-centos7-haswell:/opt/app/spack/share/spack/modules/linux-centos7-cascadelake:/usr/share/Modules/modulefiles:/etc/modulefiles:/opt/app/modulefiles"
|
| 10 |
+
source /users/cx_xchen/.bashrc_12.1
|
| 11 |
+
export DETECTRON2_DATASETS=/users/cx_xchen/DATASETS/
|
| 12 |
+
export TORCH_DISTRIBUTED_DEBUG=DETAIL
|
| 13 |
+
export OMP_NUM_THREADS=1
|
| 14 |
+
export USE_SIMPLE_THREADED_LEVEL3=1
|
| 15 |
+
conda activate frozenseg
|
| 16 |
+
|
| 17 |
+
configs=(
|
| 18 |
+
# "configs/coco/frozenseg/convnext_large_eval_a847.yaml"
|
| 19 |
+
# "configs/coco/frozenseg/convnext_large_eval_ade20k.yaml"
|
| 20 |
+
# "configs/coco/frozenseg/convnext_large_eval_lvis.yaml"
|
| 21 |
+
# "configs/coco/frozenseg/convnext_large_eval_pas21.yaml"
|
| 22 |
+
"configs/coco/frozenseg/convnext_large_eval_pc459.yaml"
|
| 23 |
+
# "configs/coco/frozenseg/convnext_large_eval_cityscapes.yaml"
|
| 24 |
+
# "configs/coco/frozenseg/convnext_large_eval_coco.yaml"
|
| 25 |
+
# "configs/coco/frozenseg/convnext_large_eval_mapillary_vistas.yaml"
|
| 26 |
+
# configs/coco/frozenseg/convnext_large_eval_bdd_panop.yaml
|
| 27 |
+
# configs/coco/frozenseg/convnext_large_eval_bdd_sem.yaml
|
| 28 |
+
)
|
| 29 |
+
port=$((10000 + RANDOM % 50000))
|
| 30 |
+
sam=vit_b
|
| 31 |
+
path=output/ConvNext-L_${sam}_1x
|
| 32 |
+
for config in "${configs[@]}"; do
|
| 33 |
+
python train_net.py --eval-only --num-gpus 1 --dist-url tcp://127.0.0.1:$port \
|
| 34 |
+
--config-file $config \
|
| 35 |
+
OUTPUT_DIR $path/$(basename "$config" .yaml) \
|
| 36 |
+
MODEL.WEIGHTS modified_model.pth \
|
| 37 |
+
MODEL.SAM_NAME vit_b \
|
| 38 |
+
MODEL.FROZEN_SEG.CLIP_PRETRAINED_WEIGHTS pretrained_checkpoint/models--laion--CLIP-convnext_large_d_320.laion2B-s29B-b131K-ft-soup/open_clip_pytorch_model.bin \
|
| 39 |
+
TEST.USE_SAM_MASKS False \
|
| 40 |
+
MODEL.FROZEN_SEG.GEOMETRIC_ENSEMBLE_BETA 0.6
|
| 41 |
+
done
|
| 42 |
+
|
| 43 |
+
########## with mask ensemble ########
|
| 44 |
+
# for config in "${configs[@]}"; do
|
| 45 |
+
# python train_net.py --eval-only --num-gpus 1 --dist-url tcp://127.0.0.1:$port \
|
| 46 |
+
# --config-file $config \
|
| 47 |
+
# OUTPUT_DIR $path/w_maskEnsemble/$(basename "$config" .yaml) \
|
| 48 |
+
# MODEL.WEIGHTS $path/model_final.pth \
|
| 49 |
+
# MODEL.MASK_FORMER.SAM_QUERY_FUSE_LAYER 2 \
|
| 50 |
+
# MODEL.MASK_FORMER.SAM_FEATURE_FUSE_LAYER 0 \
|
| 51 |
+
# MODEL.SAM_NAME vit_b \
|
| 52 |
+
# MODEL.FROZEN_SEG.CLIP_PRETRAINED_WEIGHTS pretrained_checkpoint/models--laion--CLIP-convnext_large_d_320.laion2B-s29B-b131K-ft-soup/open_clip_pytorch_model.bin \
|
| 53 |
+
# TEST.USE_SAM_MASKS True \
|
| 54 |
+
# TEST.PKL_SAM_MODEL_NAME vit_h
|
| 55 |
+
# done
|
| 56 |
+
|
| 57 |
+
|
| 58 |
+
########### test recall ############
|
| 59 |
+
# path=output/Sam_query/ConvNext-L_vit_b_1x
|
| 60 |
+
# for config in "${configs[@]}"; do
|
| 61 |
+
# srun python train_net.py --eval-only --num-gpus 4 --dist-url tcp://127.0.0.1:$port \
|
| 62 |
+
# --config-file $config \
|
| 63 |
+
# OUTPUT_DIR "output/Ablation/recall_withEverything/$(basename "$config" .yaml)" \
|
| 64 |
+
# MODEL.WEIGHTS "$path/model_final.pth" \
|
| 65 |
+
# TEST.USE_SAM_MASKS True \
|
| 66 |
+
# MODEL.MASK_FORMER.TEST.RECALL_ON True \
|
| 67 |
+
# MODEL.MASK_FORMER.TEST.SEMANTIC_ON False \
|
| 68 |
+
# MODEL.MASK_FORMER.TEST.INSTANCE_ON False \
|
| 69 |
+
# MODEL.MASK_FORMER.TEST.PANOPTIC_ON False \
|
| 70 |
+
# done
|
frozenseg/.DS_Store
ADDED
|
Binary file (6.15 kB). View file
|
|
|
frozenseg/__init__.py
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
from . import data # register all new datasets
|
| 3 |
+
from . import modeling
|
| 4 |
+
|
| 5 |
+
# config
|
| 6 |
+
from .config import add_maskformer2_config, add_frozenseg_config
|
| 7 |
+
|
| 8 |
+
# dataset loading
|
| 9 |
+
from .data.dataset_mappers.coco_instance_new_baseline_dataset_mapper import COCOInstanceNewBaselineDatasetMapper
|
| 10 |
+
from .data.dataset_mappers.coco_panoptic_new_baseline_dataset_mapper import COCOPanopticNewBaselineDatasetMapper
|
| 11 |
+
from .data.dataset_mappers.mask_former_instance_dataset_mapper import (
|
| 12 |
+
MaskFormerInstanceDatasetMapper,
|
| 13 |
+
)
|
| 14 |
+
from .data.dataset_mappers.mask_former_panoptic_dataset_mapper import (
|
| 15 |
+
MaskFormerPanopticDatasetMapper,
|
| 16 |
+
)
|
| 17 |
+
from .data.dataset_mappers.mask_former_semantic_dataset_mapper import (
|
| 18 |
+
MaskFormerSemanticDatasetMapper,
|
| 19 |
+
)
|
| 20 |
+
|
| 21 |
+
# models
|
| 22 |
+
from .frozenseg import FrozenSeg
|
| 23 |
+
from .test_time_augmentation import SemanticSegmentorWithTTA
|
| 24 |
+
|
| 25 |
+
# evaluation
|
| 26 |
+
from .evaluation.instance_evaluation import InstanceSegEvaluator
|
frozenseg/config.py
ADDED
|
@@ -0,0 +1,132 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# -*- coding: utf-8 -*-
|
| 2 |
+
from detectron2.config import CfgNode as CN
|
| 3 |
+
|
| 4 |
+
def add_maskformer2_config(cfg):
|
| 5 |
+
"""
|
| 6 |
+
Add config for MASK_FORMER.
|
| 7 |
+
"""
|
| 8 |
+
# NOTE: configs from original maskformer
|
| 9 |
+
# data config
|
| 10 |
+
# select the dataset mapper
|
| 11 |
+
cfg.INPUT.DATASET_MAPPER_NAME = "mask_former_semantic"
|
| 12 |
+
# Color augmentation
|
| 13 |
+
cfg.INPUT.COLOR_AUG_SSD = False
|
| 14 |
+
# We retry random cropping until no single category in semantic segmentation GT occupies more
|
| 15 |
+
# than `SINGLE_CATEGORY_MAX_AREA` part of the crop.
|
| 16 |
+
cfg.INPUT.CROP.SINGLE_CATEGORY_MAX_AREA = 1.0
|
| 17 |
+
# Pad image and segmentation GT in dataset mapper.
|
| 18 |
+
cfg.INPUT.SIZE_DIVISIBILITY = -1
|
| 19 |
+
|
| 20 |
+
# solver config
|
| 21 |
+
# weight decay on embedding
|
| 22 |
+
cfg.SOLVER.WEIGHT_DECAY_EMBED = 0.0
|
| 23 |
+
# optimizer
|
| 24 |
+
cfg.SOLVER.OPTIMIZER = "ADAMW"
|
| 25 |
+
cfg.SOLVER.BACKBONE_MULTIPLIER = 0.1
|
| 26 |
+
|
| 27 |
+
# mask_former model config
|
| 28 |
+
cfg.MODEL.MASK_FORMER = CN()
|
| 29 |
+
|
| 30 |
+
# loss
|
| 31 |
+
cfg.MODEL.MASK_FORMER.DEEP_SUPERVISION = True
|
| 32 |
+
cfg.MODEL.MASK_FORMER.NO_OBJECT_WEIGHT = 0.1
|
| 33 |
+
cfg.MODEL.MASK_FORMER.CLASS_WEIGHT = 1.0
|
| 34 |
+
cfg.MODEL.MASK_FORMER.DICE_WEIGHT = 1.0
|
| 35 |
+
cfg.MODEL.MASK_FORMER.MASK_WEIGHT = 20.0
|
| 36 |
+
|
| 37 |
+
# transformer config
|
| 38 |
+
cfg.MODEL.MASK_FORMER.NHEADS = 8
|
| 39 |
+
cfg.MODEL.MASK_FORMER.DROPOUT = 0.1
|
| 40 |
+
cfg.MODEL.MASK_FORMER.DIM_FEEDFORWARD = 2048
|
| 41 |
+
cfg.MODEL.MASK_FORMER.ENC_LAYERS = 0
|
| 42 |
+
cfg.MODEL.MASK_FORMER.DEC_LAYERS = 6
|
| 43 |
+
cfg.MODEL.MASK_FORMER.PRE_NORM = False
|
| 44 |
+
|
| 45 |
+
cfg.MODEL.MASK_FORMER.HIDDEN_DIM = 256
|
| 46 |
+
cfg.MODEL.MASK_FORMER.NUM_OBJECT_QUERIES = 100
|
| 47 |
+
|
| 48 |
+
cfg.MODEL.MASK_FORMER.TRANSFORMER_IN_FEATURE = "res5"
|
| 49 |
+
cfg.MODEL.MASK_FORMER.ENFORCE_INPUT_PROJ = False
|
| 50 |
+
|
| 51 |
+
# mask_former inference config
|
| 52 |
+
cfg.MODEL.MASK_FORMER.TEST = CN()
|
| 53 |
+
cfg.MODEL.MASK_FORMER.TEST.SEMANTIC_ON = True
|
| 54 |
+
cfg.MODEL.MASK_FORMER.TEST.INSTANCE_ON = False
|
| 55 |
+
cfg.MODEL.MASK_FORMER.TEST.PANOPTIC_ON = False
|
| 56 |
+
cfg.MODEL.MASK_FORMER.TEST.OBJECT_MASK_THRESHOLD = 0.0
|
| 57 |
+
cfg.MODEL.MASK_FORMER.TEST.OVERLAP_THRESHOLD = 0.0
|
| 58 |
+
cfg.MODEL.MASK_FORMER.TEST.SEM_SEG_POSTPROCESSING_BEFORE_INFERENCE = False
|
| 59 |
+
|
| 60 |
+
# Sometimes `backbone.size_divisibility` is set to 0 for some backbone (e.g. ResNet)
|
| 61 |
+
# you can use this config to override
|
| 62 |
+
cfg.MODEL.MASK_FORMER.SIZE_DIVISIBILITY = 32
|
| 63 |
+
|
| 64 |
+
# pixel decoder config
|
| 65 |
+
cfg.MODEL.SEM_SEG_HEAD.MASK_DIM = 256
|
| 66 |
+
# adding transformer in pixel decoder
|
| 67 |
+
cfg.MODEL.SEM_SEG_HEAD.TRANSFORMER_ENC_LAYERS = 0
|
| 68 |
+
# pixel decoder
|
| 69 |
+
cfg.MODEL.SEM_SEG_HEAD.PIXEL_DECODER_NAME = "BasePixelDecoder"
|
| 70 |
+
|
| 71 |
+
# swin transformer backbone
|
| 72 |
+
cfg.MODEL.SWIN = CN()
|
| 73 |
+
cfg.MODEL.SWIN.PRETRAIN_IMG_SIZE = 224
|
| 74 |
+
cfg.MODEL.SWIN.PATCH_SIZE = 4
|
| 75 |
+
cfg.MODEL.SWIN.EMBED_DIM = 96
|
| 76 |
+
cfg.MODEL.SWIN.DEPTHS = [2, 2, 6, 2]
|
| 77 |
+
cfg.MODEL.SWIN.NUM_HEADS = [3, 6, 12, 24]
|
| 78 |
+
cfg.MODEL.SWIN.WINDOW_SIZE = 7
|
| 79 |
+
cfg.MODEL.SWIN.MLP_RATIO = 4.0
|
| 80 |
+
cfg.MODEL.SWIN.QKV_BIAS = True
|
| 81 |
+
cfg.MODEL.SWIN.QK_SCALE = None
|
| 82 |
+
cfg.MODEL.SWIN.DROP_RATE = 0.0
|
| 83 |
+
cfg.MODEL.SWIN.ATTN_DROP_RATE = 0.0
|
| 84 |
+
cfg.MODEL.SWIN.DROP_PATH_RATE = 0.3
|
| 85 |
+
cfg.MODEL.SWIN.APE = False
|
| 86 |
+
cfg.MODEL.SWIN.PATCH_NORM = True
|
| 87 |
+
cfg.MODEL.SWIN.OUT_FEATURES = ["res2", "res3", "res4", "res5"]
|
| 88 |
+
cfg.MODEL.SWIN.USE_CHECKPOINT = False
|
| 89 |
+
|
| 90 |
+
# NOTE: maskformer2 extra configs
|
| 91 |
+
# transformer module
|
| 92 |
+
cfg.MODEL.MASK_FORMER.TRANSFORMER_DECODER_NAME = "MultiScaleMaskedTransformerDecoder"
|
| 93 |
+
|
| 94 |
+
# LSJ aug
|
| 95 |
+
cfg.INPUT.IMAGE_SIZE = 1024
|
| 96 |
+
cfg.INPUT.MIN_SCALE = 0.1
|
| 97 |
+
cfg.INPUT.MAX_SCALE = 2.0
|
| 98 |
+
|
| 99 |
+
# MSDeformAttn encoder configs
|
| 100 |
+
cfg.MODEL.SEM_SEG_HEAD.DEFORMABLE_TRANSFORMER_ENCODER_IN_FEATURES = ["res3", "res4", "res5"]
|
| 101 |
+
cfg.MODEL.SEM_SEG_HEAD.DEFORMABLE_TRANSFORMER_ENCODER_N_POINTS = 4
|
| 102 |
+
cfg.MODEL.SEM_SEG_HEAD.DEFORMABLE_TRANSFORMER_ENCODER_N_HEADS = 8
|
| 103 |
+
|
| 104 |
+
# point loss configs
|
| 105 |
+
# Number of points sampled during training for a mask point head.
|
| 106 |
+
cfg.MODEL.MASK_FORMER.TRAIN_NUM_POINTS = 112 * 112
|
| 107 |
+
# Oversampling parameter for PointRend point sampling during training. Parameter `k` in the
|
| 108 |
+
# original paper.
|
| 109 |
+
cfg.MODEL.MASK_FORMER.OVERSAMPLE_RATIO = 3.0
|
| 110 |
+
# Importance sampling parameter for PointRend point sampling during training. Parametr `beta` in
|
| 111 |
+
# the original paper.
|
| 112 |
+
cfg.MODEL.MASK_FORMER.IMPORTANCE_SAMPLE_RATIO = 0.75
|
| 113 |
+
|
| 114 |
+
|
| 115 |
+
def add_frozenseg_config(cfg):
|
| 116 |
+
cfg.MODEL.SAM_NAME = 'vit_b'
|
| 117 |
+
cfg.MODEL.MASK_FORMER.SAM_QUERY_FUSE_LAYER = 2
|
| 118 |
+
cfg.MODEL.MASK_FORMER.SAM_FEATURE_FUSE_LAYER = 0
|
| 119 |
+
cfg.MODEL.MASK_FORMER.TEST.RECALL_ON = False
|
| 120 |
+
|
| 121 |
+
cfg.TEST.SAM_MASK_PRED_ALPHA = 0.2
|
| 122 |
+
cfg.TEST.USE_SAM_MASKS = False
|
| 123 |
+
cfg.TEST.PKL_SAM_MODEL_NAME = 'vit_h'
|
| 124 |
+
|
| 125 |
+
cfg.MODEL.FROZEN_SEG = CN()
|
| 126 |
+
cfg.MODEL.FROZEN_SEG.CLIP_PRETRAINED_WEIGHTS = "laion2b_s29b_b131k_ft_soup"
|
| 127 |
+
cfg.MODEL.FROZEN_SEG.CLIP_MODEL_NAME = "convnext_large_d_320"
|
| 128 |
+
cfg.MODEL.FROZEN_SEG.EMBED_DIM = 768
|
| 129 |
+
cfg.MODEL.FROZEN_SEG.GEOMETRIC_ENSEMBLE_ALPHA = 0.4
|
| 130 |
+
cfg.MODEL.FROZEN_SEG.GEOMETRIC_ENSEMBLE_BETA = 0.8
|
| 131 |
+
cfg.MODEL.FROZEN_SEG.ENSEMBLE_ON_VALID_MASK = False
|
| 132 |
+
|
frozenseg/data/.DS_Store
ADDED
|
Binary file (6.15 kB). View file
|
|
|
frozenseg/data/__init__.py
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
from . import datasets
|
frozenseg/data/dataset_mappers/__init__.py
ADDED
|
File without changes
|
frozenseg/data/dataset_mappers/bdd_semseg_dataset_mapper.py
ADDED
|
@@ -0,0 +1,107 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# --------------------------------------------------------
|
| 2 |
+
# X-Decoder -- Generalized Decoding for Pixel, Image, and Language
|
| 3 |
+
# Copyright (c) 2022 Microsoft
|
| 4 |
+
# Licensed under The MIT License [see LICENSE for details]
|
| 5 |
+
# Modified by Xueyan Zou ([email protected])
|
| 6 |
+
# --------------------------------------------------------
|
| 7 |
+
# Copyright (c) Facebook, Inc. and its affiliates.
|
| 8 |
+
import copy
|
| 9 |
+
|
| 10 |
+
import scipy.io
|
| 11 |
+
import numpy as np
|
| 12 |
+
import torch
|
| 13 |
+
from PIL import Image
|
| 14 |
+
|
| 15 |
+
from torchvision import transforms
|
| 16 |
+
from detectron2.utils import configurable
|
| 17 |
+
|
| 18 |
+
__all__ = ["BDDSemDatasetMapper"]
|
| 19 |
+
|
| 20 |
+
|
| 21 |
+
# This is specifically designed for the COCO dataset.
|
| 22 |
+
class BDDSemDatasetMapper:
|
| 23 |
+
"""
|
| 24 |
+
A callable which takes a dataset dict in Detectron2 Dataset format,
|
| 25 |
+
and map it into a format used by MaskFormer.
|
| 26 |
+
|
| 27 |
+
This dataset mapper applies the same transformation as DETR for COCO panoptic segmentation.
|
| 28 |
+
|
| 29 |
+
The callable currently does the following:
|
| 30 |
+
|
| 31 |
+
1. Read the image from "file_name"
|
| 32 |
+
2. Applies geometric transforms to the image and annotation
|
| 33 |
+
3. Find and applies suitable cropping to the image and annotation
|
| 34 |
+
4. Prepare image and annotation to Tensors
|
| 35 |
+
"""
|
| 36 |
+
|
| 37 |
+
@configurable
|
| 38 |
+
def __init__(
|
| 39 |
+
self,
|
| 40 |
+
is_train=True,
|
| 41 |
+
min_size_test=None,
|
| 42 |
+
max_size_test=None,
|
| 43 |
+
mean=None,
|
| 44 |
+
std=None,
|
| 45 |
+
):
|
| 46 |
+
"""
|
| 47 |
+
NOTE: this interface is experimental.
|
| 48 |
+
Args:
|
| 49 |
+
is_train: for training or inference
|
| 50 |
+
augmentations: a list of augmentations or deterministic transforms to apply
|
| 51 |
+
tfm_gens: data augmentation
|
| 52 |
+
image_format: an image format supported by :func:`detection_utils.read_image`.
|
| 53 |
+
"""
|
| 54 |
+
self.is_train = is_train
|
| 55 |
+
self.min_size_test = min_size_test
|
| 56 |
+
self.max_size_test = max_size_test
|
| 57 |
+
self.pixel_mean = torch.tensor(mean)[:,None,None]
|
| 58 |
+
self.pixel_std = torch.tensor(std)[:,None,None]
|
| 59 |
+
|
| 60 |
+
t = []
|
| 61 |
+
t.append(transforms.Resize(self.min_size_test, interpolation=Image.BICUBIC))
|
| 62 |
+
self.transform = transforms.Compose(t)
|
| 63 |
+
|
| 64 |
+
@classmethod
|
| 65 |
+
def from_config(cls, cfg, is_train=True):
|
| 66 |
+
ret = {
|
| 67 |
+
"is_train": is_train,
|
| 68 |
+
"min_size_test": cfg['INPUT']['MIN_SIZE_TEST'],
|
| 69 |
+
"max_size_test": cfg['INPUT']['MAX_SIZE_TEST'],
|
| 70 |
+
"mean": cfg['INPUT']['PIXEL_MEAN'],
|
| 71 |
+
"std": cfg['INPUT']['PIXEL_STD'],
|
| 72 |
+
}
|
| 73 |
+
return ret
|
| 74 |
+
|
| 75 |
+
def read_semseg(self, file_name):
|
| 76 |
+
if '.png' in file_name:
|
| 77 |
+
semseg = np.asarray(Image.open(file_name))
|
| 78 |
+
elif '.mat' in file_name:
|
| 79 |
+
semseg = scipy.io.loadmat(file_name)['LabelMap']
|
| 80 |
+
return semseg
|
| 81 |
+
|
| 82 |
+
def __call__(self, dataset_dict):
|
| 83 |
+
"""
|
| 84 |
+
Args:
|
| 85 |
+
dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format.
|
| 86 |
+
|
| 87 |
+
Returns:
|
| 88 |
+
dict: a format that builtin models in detectron2 accept
|
| 89 |
+
"""
|
| 90 |
+
dataset_dict = copy.deepcopy(dataset_dict) # it will be modified by code below
|
| 91 |
+
file_name = dataset_dict['file_name']
|
| 92 |
+
semseg_name = dataset_dict['sem_seg_file_name']
|
| 93 |
+
image = Image.open(file_name).convert('RGB')
|
| 94 |
+
|
| 95 |
+
dataset_dict['width'] = image.size[0]
|
| 96 |
+
dataset_dict['height'] = image.size[1]
|
| 97 |
+
|
| 98 |
+
if self.is_train == False:
|
| 99 |
+
image = self.transform(image)
|
| 100 |
+
image = torch.from_numpy(np.asarray(image).copy())
|
| 101 |
+
image = image.permute(2,0,1)
|
| 102 |
+
|
| 103 |
+
semseg = self.read_semseg(semseg_name)
|
| 104 |
+
semseg = torch.from_numpy(semseg.astype(np.int32))
|
| 105 |
+
dataset_dict['image'] = image
|
| 106 |
+
dataset_dict['semseg'] = semseg
|
| 107 |
+
return dataset_dict
|
frozenseg/data/dataset_mappers/coco_instance_new_baseline_dataset_mapper.py
ADDED
|
@@ -0,0 +1,187 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import copy
|
| 2 |
+
import logging
|
| 3 |
+
|
| 4 |
+
import numpy as np
|
| 5 |
+
import torch
|
| 6 |
+
|
| 7 |
+
from detectron2.config import configurable
|
| 8 |
+
from detectron2.data import detection_utils as utils
|
| 9 |
+
from detectron2.data import transforms as T
|
| 10 |
+
from detectron2.data.transforms import TransformGen
|
| 11 |
+
from detectron2.structures import BitMasks, Instances
|
| 12 |
+
|
| 13 |
+
from pycocotools import mask as coco_mask
|
| 14 |
+
|
| 15 |
+
__all__ = ["COCOInstanceNewBaselineDatasetMapper"]
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
def convert_coco_poly_to_mask(segmentations, height, width):
|
| 19 |
+
masks = []
|
| 20 |
+
for polygons in segmentations:
|
| 21 |
+
rles = coco_mask.frPyObjects(polygons, height, width)
|
| 22 |
+
mask = coco_mask.decode(rles)
|
| 23 |
+
if len(mask.shape) < 3:
|
| 24 |
+
mask = mask[..., None]
|
| 25 |
+
mask = torch.as_tensor(mask, dtype=torch.uint8)
|
| 26 |
+
mask = mask.any(dim=2)
|
| 27 |
+
masks.append(mask)
|
| 28 |
+
if masks:
|
| 29 |
+
masks = torch.stack(masks, dim=0)
|
| 30 |
+
else:
|
| 31 |
+
masks = torch.zeros((0, height, width), dtype=torch.uint8)
|
| 32 |
+
return masks
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
def build_transform_gen(cfg, is_train):
|
| 36 |
+
"""
|
| 37 |
+
Create a list of default :class:`Augmentation` from config.
|
| 38 |
+
Now it includes resizing and flipping.
|
| 39 |
+
Returns:
|
| 40 |
+
list[Augmentation]
|
| 41 |
+
"""
|
| 42 |
+
assert is_train, "Only support training augmentation"
|
| 43 |
+
image_size = cfg.INPUT.IMAGE_SIZE
|
| 44 |
+
min_scale = cfg.INPUT.MIN_SCALE
|
| 45 |
+
max_scale = cfg.INPUT.MAX_SCALE
|
| 46 |
+
|
| 47 |
+
augmentation = []
|
| 48 |
+
|
| 49 |
+
if cfg.INPUT.RANDOM_FLIP != "none":
|
| 50 |
+
augmentation.append(
|
| 51 |
+
T.RandomFlip(
|
| 52 |
+
horizontal=cfg.INPUT.RANDOM_FLIP == "horizontal",
|
| 53 |
+
vertical=cfg.INPUT.RANDOM_FLIP == "vertical",
|
| 54 |
+
)
|
| 55 |
+
)
|
| 56 |
+
|
| 57 |
+
augmentation.extend([
|
| 58 |
+
T.ResizeScale(
|
| 59 |
+
min_scale=min_scale, max_scale=max_scale, target_height=image_size, target_width=image_size
|
| 60 |
+
),
|
| 61 |
+
T.FixedSizeCrop(crop_size=(image_size, image_size)),
|
| 62 |
+
])
|
| 63 |
+
|
| 64 |
+
return augmentation
|
| 65 |
+
|
| 66 |
+
|
| 67 |
+
# This is specifically designed for the COCO dataset.
|
| 68 |
+
class COCOInstanceNewBaselineDatasetMapper:
|
| 69 |
+
"""
|
| 70 |
+
A callable which takes a dataset dict in Detectron2 Dataset format,
|
| 71 |
+
and map it into a format used by MaskFormer.
|
| 72 |
+
|
| 73 |
+
This dataset mapper applies the same transformation as DETR for COCO panoptic segmentation.
|
| 74 |
+
|
| 75 |
+
The callable currently does the following:
|
| 76 |
+
|
| 77 |
+
1. Read the image from "file_name"
|
| 78 |
+
2. Applies geometric transforms to the image and annotation
|
| 79 |
+
3. Find and applies suitable cropping to the image and annotation
|
| 80 |
+
4. Prepare image and annotation to Tensors
|
| 81 |
+
"""
|
| 82 |
+
|
| 83 |
+
@configurable
|
| 84 |
+
def __init__(
|
| 85 |
+
self,
|
| 86 |
+
is_train=True,
|
| 87 |
+
*,
|
| 88 |
+
tfm_gens,
|
| 89 |
+
image_format,
|
| 90 |
+
):
|
| 91 |
+
"""
|
| 92 |
+
NOTE: this interface is experimental.
|
| 93 |
+
Args:
|
| 94 |
+
is_train: for training or inference
|
| 95 |
+
augmentations: a list of augmentations or deterministic transforms to apply
|
| 96 |
+
tfm_gens: data augmentation
|
| 97 |
+
image_format: an image format supported by :func:`detection_utils.read_image`.
|
| 98 |
+
"""
|
| 99 |
+
self.tfm_gens = tfm_gens
|
| 100 |
+
logging.getLogger(__name__).info(
|
| 101 |
+
"[COCOInstanceNewBaselineDatasetMapper] Full TransformGens used in training: {}".format(str(self.tfm_gens))
|
| 102 |
+
)
|
| 103 |
+
|
| 104 |
+
self.img_format = image_format
|
| 105 |
+
self.is_train = is_train
|
| 106 |
+
|
| 107 |
+
@classmethod
|
| 108 |
+
def from_config(cls, cfg, is_train=True):
|
| 109 |
+
# Build augmentation
|
| 110 |
+
tfm_gens = build_transform_gen(cfg, is_train)
|
| 111 |
+
|
| 112 |
+
ret = {
|
| 113 |
+
"is_train": is_train,
|
| 114 |
+
"tfm_gens": tfm_gens,
|
| 115 |
+
"image_format": cfg.INPUT.FORMAT,
|
| 116 |
+
}
|
| 117 |
+
return ret
|
| 118 |
+
|
| 119 |
+
def __call__(self, dataset_dict):
|
| 120 |
+
"""
|
| 121 |
+
Args:
|
| 122 |
+
dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format.
|
| 123 |
+
|
| 124 |
+
Returns:
|
| 125 |
+
dict: a format that builtin models in detectron2 accept
|
| 126 |
+
"""
|
| 127 |
+
dataset_dict = copy.deepcopy(dataset_dict) # it will be modified by code below
|
| 128 |
+
image = utils.read_image(dataset_dict["file_name"], format=self.img_format)
|
| 129 |
+
utils.check_image_size(dataset_dict, image)
|
| 130 |
+
|
| 131 |
+
# TODO: get padding mask
|
| 132 |
+
# by feeding a "segmentation mask" to the same transforms
|
| 133 |
+
padding_mask = np.ones(image.shape[:2])
|
| 134 |
+
|
| 135 |
+
image, transforms = T.apply_transform_gens(self.tfm_gens, image)
|
| 136 |
+
# the crop transformation has default padding value 0 for segmentation
|
| 137 |
+
padding_mask = transforms.apply_segmentation(padding_mask)
|
| 138 |
+
padding_mask = ~ padding_mask.astype(bool)
|
| 139 |
+
|
| 140 |
+
image_shape = image.shape[:2] # h, w
|
| 141 |
+
|
| 142 |
+
# Pytorch's dataloader is efficient on torch.Tensor due to shared-memory,
|
| 143 |
+
# but not efficient on large generic data structures due to the use of pickle & mp.Queue.
|
| 144 |
+
# Therefore it's important to use torch.Tensor.
|
| 145 |
+
dataset_dict["image"] = torch.as_tensor(np.ascontiguousarray(image.transpose(2, 0, 1)))
|
| 146 |
+
dataset_dict["padding_mask"] = torch.as_tensor(np.ascontiguousarray(padding_mask))
|
| 147 |
+
|
| 148 |
+
if not self.is_train:
|
| 149 |
+
# USER: Modify this if you want to keep them for some reason.
|
| 150 |
+
dataset_dict.pop("annotations", None)
|
| 151 |
+
return dataset_dict
|
| 152 |
+
|
| 153 |
+
if "annotations" in dataset_dict:
|
| 154 |
+
# USER: Modify this if you want to keep them for some reason.
|
| 155 |
+
for anno in dataset_dict["annotations"]:
|
| 156 |
+
# Let's always keep mask
|
| 157 |
+
# if not self.mask_on:
|
| 158 |
+
# anno.pop("segmentation", None)
|
| 159 |
+
anno.pop("keypoints", None)
|
| 160 |
+
|
| 161 |
+
# USER: Implement additional transformations if you have other types of data
|
| 162 |
+
annos = [
|
| 163 |
+
utils.transform_instance_annotations(obj, transforms, image_shape)
|
| 164 |
+
for obj in dataset_dict.pop("annotations")
|
| 165 |
+
if obj.get("iscrowd", 0) == 0
|
| 166 |
+
]
|
| 167 |
+
# NOTE: does not support BitMask due to augmentation
|
| 168 |
+
# Current BitMask cannot handle empty objects
|
| 169 |
+
instances = utils.annotations_to_instances(annos, image_shape)
|
| 170 |
+
# After transforms such as cropping are applied, the bounding box may no longer
|
| 171 |
+
# tightly bound the object. As an example, imagine a triangle object
|
| 172 |
+
# [(0,0), (2,0), (0,2)] cropped by a box [(1,0),(2,2)] (XYXY format). The tight
|
| 173 |
+
# bounding box of the cropped triangle should be [(1,0),(2,1)], which is not equal to
|
| 174 |
+
# the intersection of original bounding box and the cropping box.
|
| 175 |
+
instances.gt_boxes = instances.gt_masks.get_bounding_boxes()
|
| 176 |
+
# Need to filter empty instances first (due to augmentation)
|
| 177 |
+
instances = utils.filter_empty_instances(instances)
|
| 178 |
+
# Generate masks from polygon
|
| 179 |
+
h, w = instances.image_size
|
| 180 |
+
# image_size_xyxy = torch.as_tensor([w, h, w, h], dtype=torch.float)
|
| 181 |
+
if hasattr(instances, 'gt_masks'):
|
| 182 |
+
gt_masks = instances.gt_masks
|
| 183 |
+
gt_masks = convert_coco_poly_to_mask(gt_masks.polygons, h, w)
|
| 184 |
+
instances.gt_masks = gt_masks
|
| 185 |
+
dataset_dict["instances"] = instances
|
| 186 |
+
|
| 187 |
+
return dataset_dict
|
frozenseg/data/dataset_mappers/coco_panoptic_new_baseline_dataset_mapper.py
ADDED
|
@@ -0,0 +1,163 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import copy
|
| 2 |
+
import logging
|
| 3 |
+
|
| 4 |
+
import numpy as np
|
| 5 |
+
import torch
|
| 6 |
+
|
| 7 |
+
from detectron2.config import configurable
|
| 8 |
+
from detectron2.data import detection_utils as utils
|
| 9 |
+
from detectron2.data import transforms as T
|
| 10 |
+
from detectron2.data.transforms import TransformGen
|
| 11 |
+
from detectron2.structures import BitMasks, Boxes, Instances
|
| 12 |
+
|
| 13 |
+
__all__ = ["COCOPanopticNewBaselineDatasetMapper"]
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
def build_transform_gen(cfg, is_train):
|
| 17 |
+
"""
|
| 18 |
+
Create a list of default :class:`Augmentation` from config.
|
| 19 |
+
Now it includes resizing and flipping.
|
| 20 |
+
Returns:
|
| 21 |
+
list[Augmentation]
|
| 22 |
+
"""
|
| 23 |
+
assert is_train, "Only support training augmentation"
|
| 24 |
+
image_size = cfg.INPUT.IMAGE_SIZE
|
| 25 |
+
min_scale = cfg.INPUT.MIN_SCALE
|
| 26 |
+
max_scale = cfg.INPUT.MAX_SCALE
|
| 27 |
+
|
| 28 |
+
augmentation = []
|
| 29 |
+
|
| 30 |
+
if cfg.INPUT.RANDOM_FLIP != "none":
|
| 31 |
+
augmentation.append(
|
| 32 |
+
T.RandomFlip(
|
| 33 |
+
horizontal=cfg.INPUT.RANDOM_FLIP == "horizontal",
|
| 34 |
+
vertical=cfg.INPUT.RANDOM_FLIP == "vertical",
|
| 35 |
+
)
|
| 36 |
+
)
|
| 37 |
+
|
| 38 |
+
augmentation.extend([
|
| 39 |
+
T.ResizeScale(
|
| 40 |
+
min_scale=min_scale, max_scale=max_scale, target_height=image_size, target_width=image_size
|
| 41 |
+
),
|
| 42 |
+
T.FixedSizeCrop(crop_size=(image_size, image_size)),
|
| 43 |
+
])
|
| 44 |
+
|
| 45 |
+
return augmentation
|
| 46 |
+
|
| 47 |
+
|
| 48 |
+
# This is specifically designed for the COCO dataset.
|
| 49 |
+
class COCOPanopticNewBaselineDatasetMapper:
|
| 50 |
+
"""
|
| 51 |
+
A callable which takes a dataset dict in Detectron2 Dataset format,
|
| 52 |
+
and map it into a format used by MaskFormer.
|
| 53 |
+
|
| 54 |
+
This dataset mapper applies the same transformation as DETR for COCO panoptic segmentation.
|
| 55 |
+
|
| 56 |
+
The callable currently does the following:
|
| 57 |
+
|
| 58 |
+
1. Read the image from "file_name"
|
| 59 |
+
2. Applies geometric transforms to the image and annotation
|
| 60 |
+
3. Find and applies suitable cropping to the image and annotation
|
| 61 |
+
4. Prepare image and annotation to Tensors
|
| 62 |
+
"""
|
| 63 |
+
|
| 64 |
+
@configurable
|
| 65 |
+
def __init__(
|
| 66 |
+
self,
|
| 67 |
+
is_train=True,
|
| 68 |
+
*,
|
| 69 |
+
tfm_gens,
|
| 70 |
+
image_format,
|
| 71 |
+
):
|
| 72 |
+
"""
|
| 73 |
+
NOTE: this interface is experimental.
|
| 74 |
+
Args:
|
| 75 |
+
is_train: for training or inference
|
| 76 |
+
augmentations: a list of augmentations or deterministic transforms to apply
|
| 77 |
+
crop_gen: crop augmentation
|
| 78 |
+
tfm_gens: data augmentation
|
| 79 |
+
image_format: an image format supported by :func:`detection_utils.read_image`.
|
| 80 |
+
"""
|
| 81 |
+
self.tfm_gens = tfm_gens
|
| 82 |
+
logging.getLogger(__name__).info(
|
| 83 |
+
"[COCOPanopticNewBaselineDatasetMapper] Full TransformGens used in training: {}".format(
|
| 84 |
+
str(self.tfm_gens)
|
| 85 |
+
)
|
| 86 |
+
)
|
| 87 |
+
|
| 88 |
+
self.img_format = image_format
|
| 89 |
+
self.is_train = is_train
|
| 90 |
+
|
| 91 |
+
@classmethod
|
| 92 |
+
def from_config(cls, cfg, is_train=True):
|
| 93 |
+
# Build augmentation
|
| 94 |
+
tfm_gens = build_transform_gen(cfg, is_train)
|
| 95 |
+
|
| 96 |
+
ret = {
|
| 97 |
+
"is_train": is_train,
|
| 98 |
+
"tfm_gens": tfm_gens,
|
| 99 |
+
"image_format": cfg.INPUT.FORMAT,
|
| 100 |
+
}
|
| 101 |
+
return ret
|
| 102 |
+
|
| 103 |
+
def __call__(self, dataset_dict):
|
| 104 |
+
"""
|
| 105 |
+
Args:
|
| 106 |
+
dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format.
|
| 107 |
+
|
| 108 |
+
Returns:
|
| 109 |
+
dict: a format that builtin models in detectron2 accept
|
| 110 |
+
"""
|
| 111 |
+
dataset_dict = copy.deepcopy(dataset_dict) # it will be modified by code below
|
| 112 |
+
image = utils.read_image(dataset_dict["file_name"], format=self.img_format)
|
| 113 |
+
utils.check_image_size(dataset_dict, image)
|
| 114 |
+
|
| 115 |
+
image, transforms = T.apply_transform_gens(self.tfm_gens, image)
|
| 116 |
+
image_shape = image.shape[:2] # h, w
|
| 117 |
+
|
| 118 |
+
# Pytorch's dataloader is efficient on torch.Tensor due to shared-memory,
|
| 119 |
+
# but not efficient on large generic data structures due to the use of pickle & mp.Queue.
|
| 120 |
+
# Therefore it's important to use torch.Tensor.
|
| 121 |
+
dataset_dict["image"] = torch.as_tensor(np.ascontiguousarray(image.transpose(2, 0, 1)))
|
| 122 |
+
|
| 123 |
+
if not self.is_train:
|
| 124 |
+
# USER: Modify this if you want to keep them for some reason.
|
| 125 |
+
dataset_dict.pop("annotations", None)
|
| 126 |
+
return dataset_dict
|
| 127 |
+
|
| 128 |
+
if "pan_seg_file_name" in dataset_dict:
|
| 129 |
+
pan_seg_gt = utils.read_image(dataset_dict.pop("pan_seg_file_name"), "RGB")
|
| 130 |
+
segments_info = dataset_dict["segments_info"]
|
| 131 |
+
|
| 132 |
+
# apply the same transformation to panoptic segmentation
|
| 133 |
+
pan_seg_gt = transforms.apply_segmentation(pan_seg_gt)
|
| 134 |
+
|
| 135 |
+
from panopticapi.utils import rgb2id
|
| 136 |
+
|
| 137 |
+
pan_seg_gt = rgb2id(pan_seg_gt)
|
| 138 |
+
|
| 139 |
+
instances = Instances(image_shape)
|
| 140 |
+
classes = []
|
| 141 |
+
masks = []
|
| 142 |
+
for segment_info in segments_info:
|
| 143 |
+
class_id = segment_info["category_id"]
|
| 144 |
+
if not segment_info["iscrowd"]:
|
| 145 |
+
classes.append(class_id)
|
| 146 |
+
masks.append(pan_seg_gt == segment_info["id"])
|
| 147 |
+
|
| 148 |
+
classes = np.array(classes)
|
| 149 |
+
instances.gt_classes = torch.tensor(classes, dtype=torch.int64)
|
| 150 |
+
if len(masks) == 0:
|
| 151 |
+
# Some image does not have annotation (all ignored)
|
| 152 |
+
instances.gt_masks = torch.zeros((0, pan_seg_gt.shape[-2], pan_seg_gt.shape[-1]))
|
| 153 |
+
instances.gt_boxes = Boxes(torch.zeros((0, 4)))
|
| 154 |
+
else:
|
| 155 |
+
masks = BitMasks(
|
| 156 |
+
torch.stack([torch.from_numpy(np.ascontiguousarray(x.copy())) for x in masks])
|
| 157 |
+
)
|
| 158 |
+
instances.gt_masks = masks.tensor
|
| 159 |
+
instances.gt_boxes = masks.get_bounding_boxes()
|
| 160 |
+
|
| 161 |
+
dataset_dict["instances"] = instances
|
| 162 |
+
|
| 163 |
+
return dataset_dict
|
frozenseg/data/dataset_mappers/mask_former_instance_dataset_mapper.py
ADDED
|
@@ -0,0 +1,179 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import copy
|
| 2 |
+
import logging
|
| 3 |
+
|
| 4 |
+
import numpy as np
|
| 5 |
+
import pycocotools.mask as mask_util
|
| 6 |
+
import torch
|
| 7 |
+
from torch.nn import functional as F
|
| 8 |
+
|
| 9 |
+
from detectron2.config import configurable
|
| 10 |
+
from detectron2.data import detection_utils as utils
|
| 11 |
+
from detectron2.data import transforms as T
|
| 12 |
+
from detectron2.projects.point_rend import ColorAugSSDTransform
|
| 13 |
+
from detectron2.structures import BitMasks, Instances, polygons_to_bitmask
|
| 14 |
+
|
| 15 |
+
__all__ = ["MaskFormerInstanceDatasetMapper"]
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
class MaskFormerInstanceDatasetMapper:
|
| 19 |
+
"""
|
| 20 |
+
A callable which takes a dataset dict in Detectron2 Dataset format,
|
| 21 |
+
and map it into a format used by MaskFormer for instance segmentation.
|
| 22 |
+
|
| 23 |
+
The callable currently does the following:
|
| 24 |
+
|
| 25 |
+
1. Read the image from "file_name"
|
| 26 |
+
2. Applies geometric transforms to the image and annotation
|
| 27 |
+
3. Find and applies suitable cropping to the image and annotation
|
| 28 |
+
4. Prepare image and annotation to Tensors
|
| 29 |
+
"""
|
| 30 |
+
|
| 31 |
+
@configurable
|
| 32 |
+
def __init__(
|
| 33 |
+
self,
|
| 34 |
+
is_train=True,
|
| 35 |
+
*,
|
| 36 |
+
augmentations,
|
| 37 |
+
image_format,
|
| 38 |
+
size_divisibility,
|
| 39 |
+
):
|
| 40 |
+
"""
|
| 41 |
+
NOTE: this interface is experimental.
|
| 42 |
+
Args:
|
| 43 |
+
is_train: for training or inference
|
| 44 |
+
augmentations: a list of augmentations or deterministic transforms to apply
|
| 45 |
+
image_format: an image format supported by :func:`detection_utils.read_image`.
|
| 46 |
+
size_divisibility: pad image size to be divisible by this value
|
| 47 |
+
"""
|
| 48 |
+
self.is_train = is_train
|
| 49 |
+
self.tfm_gens = augmentations
|
| 50 |
+
self.img_format = image_format
|
| 51 |
+
self.size_divisibility = size_divisibility
|
| 52 |
+
|
| 53 |
+
logger = logging.getLogger(__name__)
|
| 54 |
+
mode = "training" if is_train else "inference"
|
| 55 |
+
logger.info(f"[{self.__class__.__name__}] Augmentations used in {mode}: {augmentations}")
|
| 56 |
+
|
| 57 |
+
@classmethod
|
| 58 |
+
def from_config(cls, cfg, is_train=True):
|
| 59 |
+
# Build augmentation
|
| 60 |
+
augs = [
|
| 61 |
+
T.ResizeShortestEdge(
|
| 62 |
+
cfg.INPUT.MIN_SIZE_TRAIN,
|
| 63 |
+
cfg.INPUT.MAX_SIZE_TRAIN,
|
| 64 |
+
cfg.INPUT.MIN_SIZE_TRAIN_SAMPLING,
|
| 65 |
+
)
|
| 66 |
+
]
|
| 67 |
+
if cfg.INPUT.CROP.ENABLED:
|
| 68 |
+
augs.append(
|
| 69 |
+
T.RandomCrop(
|
| 70 |
+
cfg.INPUT.CROP.TYPE,
|
| 71 |
+
cfg.INPUT.CROP.SIZE,
|
| 72 |
+
)
|
| 73 |
+
)
|
| 74 |
+
if cfg.INPUT.COLOR_AUG_SSD:
|
| 75 |
+
augs.append(ColorAugSSDTransform(img_format=cfg.INPUT.FORMAT))
|
| 76 |
+
augs.append(T.RandomFlip())
|
| 77 |
+
|
| 78 |
+
ret = {
|
| 79 |
+
"is_train": is_train,
|
| 80 |
+
"augmentations": augs,
|
| 81 |
+
"image_format": cfg.INPUT.FORMAT,
|
| 82 |
+
"size_divisibility": cfg.INPUT.SIZE_DIVISIBILITY,
|
| 83 |
+
}
|
| 84 |
+
return ret
|
| 85 |
+
|
| 86 |
+
def __call__(self, dataset_dict):
|
| 87 |
+
"""
|
| 88 |
+
Args:
|
| 89 |
+
dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format.
|
| 90 |
+
|
| 91 |
+
Returns:
|
| 92 |
+
dict: a format that builtin models in detectron2 accept
|
| 93 |
+
"""
|
| 94 |
+
assert self.is_train, "MaskFormerPanopticDatasetMapper should only be used for training!"
|
| 95 |
+
|
| 96 |
+
dataset_dict = copy.deepcopy(dataset_dict) # it will be modified by code below
|
| 97 |
+
image = utils.read_image(dataset_dict["file_name"], format=self.img_format)
|
| 98 |
+
utils.check_image_size(dataset_dict, image)
|
| 99 |
+
|
| 100 |
+
aug_input = T.AugInput(image)
|
| 101 |
+
aug_input, transforms = T.apply_transform_gens(self.tfm_gens, aug_input)
|
| 102 |
+
image = aug_input.image
|
| 103 |
+
|
| 104 |
+
# transform instnace masks
|
| 105 |
+
assert "annotations" in dataset_dict
|
| 106 |
+
for anno in dataset_dict["annotations"]:
|
| 107 |
+
anno.pop("keypoints", None)
|
| 108 |
+
|
| 109 |
+
annos = [
|
| 110 |
+
utils.transform_instance_annotations(obj, transforms, image.shape[:2])
|
| 111 |
+
for obj in dataset_dict.pop("annotations")
|
| 112 |
+
if obj.get("iscrowd", 0) == 0
|
| 113 |
+
]
|
| 114 |
+
|
| 115 |
+
if len(annos):
|
| 116 |
+
assert "segmentation" in annos[0]
|
| 117 |
+
segms = [obj["segmentation"] for obj in annos]
|
| 118 |
+
masks = []
|
| 119 |
+
for segm in segms:
|
| 120 |
+
if isinstance(segm, list):
|
| 121 |
+
# polygon
|
| 122 |
+
masks.append(polygons_to_bitmask(segm, *image.shape[:2]))
|
| 123 |
+
elif isinstance(segm, dict):
|
| 124 |
+
# COCO RLE
|
| 125 |
+
masks.append(mask_util.decode(segm))
|
| 126 |
+
elif isinstance(segm, np.ndarray):
|
| 127 |
+
assert segm.ndim == 2, "Expect segmentation of 2 dimensions, got {}.".format(
|
| 128 |
+
segm.ndim
|
| 129 |
+
)
|
| 130 |
+
# mask array
|
| 131 |
+
masks.append(segm)
|
| 132 |
+
else:
|
| 133 |
+
raise ValueError(
|
| 134 |
+
"Cannot convert segmentation of type '{}' to BitMasks!"
|
| 135 |
+
"Supported types are: polygons as list[list[float] or ndarray],"
|
| 136 |
+
" COCO-style RLE as a dict, or a binary segmentation mask "
|
| 137 |
+
" in a 2D numpy array of shape HxW.".format(type(segm))
|
| 138 |
+
)
|
| 139 |
+
|
| 140 |
+
# Pad image and segmentation label here!
|
| 141 |
+
image = torch.as_tensor(np.ascontiguousarray(image.transpose(2, 0, 1)))
|
| 142 |
+
masks = [torch.from_numpy(np.ascontiguousarray(x)) for x in masks]
|
| 143 |
+
|
| 144 |
+
classes = [int(obj["category_id"]) for obj in annos]
|
| 145 |
+
classes = torch.tensor(classes, dtype=torch.int64)
|
| 146 |
+
|
| 147 |
+
if self.size_divisibility > 0:
|
| 148 |
+
image_size = (image.shape[-2], image.shape[-1])
|
| 149 |
+
padding_size = [
|
| 150 |
+
0,
|
| 151 |
+
self.size_divisibility - image_size[1],
|
| 152 |
+
0,
|
| 153 |
+
self.size_divisibility - image_size[0],
|
| 154 |
+
]
|
| 155 |
+
# pad image
|
| 156 |
+
image = F.pad(image, padding_size, value=128).contiguous()
|
| 157 |
+
# pad mask
|
| 158 |
+
masks = [F.pad(x, padding_size, value=0).contiguous() for x in masks]
|
| 159 |
+
|
| 160 |
+
image_shape = (image.shape[-2], image.shape[-1]) # h, w
|
| 161 |
+
|
| 162 |
+
# Pytorch's dataloader is efficient on torch.Tensor due to shared-memory,
|
| 163 |
+
# but not efficient on large generic data structures due to the use of pickle & mp.Queue.
|
| 164 |
+
# Therefore it's important to use torch.Tensor.
|
| 165 |
+
dataset_dict["image"] = image
|
| 166 |
+
|
| 167 |
+
# Prepare per-category binary masks
|
| 168 |
+
instances = Instances(image_shape)
|
| 169 |
+
instances.gt_classes = classes
|
| 170 |
+
if len(masks) == 0:
|
| 171 |
+
# Some image does not have annotation (all ignored)
|
| 172 |
+
instances.gt_masks = torch.zeros((0, image.shape[-2], image.shape[-1]))
|
| 173 |
+
else:
|
| 174 |
+
masks = BitMasks(torch.stack(masks))
|
| 175 |
+
instances.gt_masks = masks.tensor
|
| 176 |
+
|
| 177 |
+
dataset_dict["instances"] = instances
|
| 178 |
+
|
| 179 |
+
return dataset_dict
|
frozenseg/data/dataset_mappers/mask_former_panoptic_dataset_mapper.py
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import copy
|
| 2 |
+
import logging
|
| 3 |
+
|
| 4 |
+
import numpy as np
|
| 5 |
+
import torch
|
| 6 |
+
from torch.nn import functional as F
|
| 7 |
+
|
| 8 |
+
from detectron2.config import configurable
|
| 9 |
+
from detectron2.data import detection_utils as utils
|
| 10 |
+
from detectron2.data import transforms as T
|
| 11 |
+
from detectron2.structures import BitMasks, Instances
|
| 12 |
+
|
| 13 |
+
from .mask_former_semantic_dataset_mapper import MaskFormerSemanticDatasetMapper
|
| 14 |
+
|
| 15 |
+
__all__ = ["MaskFormerPanopticDatasetMapper"]
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
class MaskFormerPanopticDatasetMapper(MaskFormerSemanticDatasetMapper):
|
| 19 |
+
"""
|
| 20 |
+
A callable which takes a dataset dict in Detectron2 Dataset format,
|
| 21 |
+
and map it into a format used by MaskFormer for panoptic segmentation.
|
| 22 |
+
|
| 23 |
+
The callable currently does the following:
|
| 24 |
+
|
| 25 |
+
1. Read the image from "file_name"
|
| 26 |
+
2. Applies geometric transforms to the image and annotation
|
| 27 |
+
3. Find and applies suitable cropping to the image and annotation
|
| 28 |
+
4. Prepare image and annotation to Tensors
|
| 29 |
+
"""
|
| 30 |
+
|
| 31 |
+
@configurable
|
| 32 |
+
def __init__(
|
| 33 |
+
self,
|
| 34 |
+
is_train=True,
|
| 35 |
+
*,
|
| 36 |
+
augmentations,
|
| 37 |
+
image_format,
|
| 38 |
+
ignore_label,
|
| 39 |
+
size_divisibility,
|
| 40 |
+
):
|
| 41 |
+
"""
|
| 42 |
+
NOTE: this interface is experimental.
|
| 43 |
+
Args:
|
| 44 |
+
is_train: for training or inference
|
| 45 |
+
augmentations: a list of augmentations or deterministic transforms to apply
|
| 46 |
+
image_format: an image format supported by :func:`detection_utils.read_image`.
|
| 47 |
+
ignore_label: the label that is ignored to evaluation
|
| 48 |
+
size_divisibility: pad image size to be divisible by this value
|
| 49 |
+
"""
|
| 50 |
+
super().__init__(
|
| 51 |
+
is_train,
|
| 52 |
+
augmentations=augmentations,
|
| 53 |
+
image_format=image_format,
|
| 54 |
+
ignore_label=ignore_label,
|
| 55 |
+
size_divisibility=size_divisibility,
|
| 56 |
+
)
|
| 57 |
+
|
| 58 |
+
def __call__(self, dataset_dict):
|
| 59 |
+
"""
|
| 60 |
+
Args:
|
| 61 |
+
dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format.
|
| 62 |
+
|
| 63 |
+
Returns:
|
| 64 |
+
dict: a format that builtin models in detectron2 accept
|
| 65 |
+
"""
|
| 66 |
+
assert self.is_train, "MaskFormerPanopticDatasetMapper should only be used for training!"
|
| 67 |
+
|
| 68 |
+
dataset_dict = copy.deepcopy(dataset_dict) # it will be modified by code below
|
| 69 |
+
image = utils.read_image(dataset_dict["file_name"], format=self.img_format)
|
| 70 |
+
utils.check_image_size(dataset_dict, image)
|
| 71 |
+
|
| 72 |
+
# semantic segmentation
|
| 73 |
+
if "sem_seg_file_name" in dataset_dict:
|
| 74 |
+
# PyTorch transformation not implemented for uint16, so converting it to double first
|
| 75 |
+
sem_seg_gt = utils.read_image(dataset_dict.pop("sem_seg_file_name")).astype("double")
|
| 76 |
+
else:
|
| 77 |
+
sem_seg_gt = None
|
| 78 |
+
|
| 79 |
+
# panoptic segmentation
|
| 80 |
+
if "pan_seg_file_name" in dataset_dict:
|
| 81 |
+
pan_seg_gt = utils.read_image(dataset_dict.pop("pan_seg_file_name"), "RGB")
|
| 82 |
+
segments_info = dataset_dict["segments_info"]
|
| 83 |
+
else:
|
| 84 |
+
pan_seg_gt = None
|
| 85 |
+
segments_info = None
|
| 86 |
+
|
| 87 |
+
if pan_seg_gt is None:
|
| 88 |
+
raise ValueError(
|
| 89 |
+
"Cannot find 'pan_seg_file_name' for panoptic segmentation dataset {}.".format(
|
| 90 |
+
dataset_dict["file_name"]
|
| 91 |
+
)
|
| 92 |
+
)
|
| 93 |
+
|
| 94 |
+
aug_input = T.AugInput(image, sem_seg=sem_seg_gt)
|
| 95 |
+
aug_input, transforms = T.apply_transform_gens(self.tfm_gens, aug_input)
|
| 96 |
+
image = aug_input.image
|
| 97 |
+
if sem_seg_gt is not None:
|
| 98 |
+
sem_seg_gt = aug_input.sem_seg
|
| 99 |
+
|
| 100 |
+
# apply the same transformation to panoptic segmentation
|
| 101 |
+
pan_seg_gt = transforms.apply_segmentation(pan_seg_gt)
|
| 102 |
+
|
| 103 |
+
from panopticapi.utils import rgb2id
|
| 104 |
+
|
| 105 |
+
pan_seg_gt = rgb2id(pan_seg_gt)
|
| 106 |
+
|
| 107 |
+
# Pad image and segmentation label here!
|
| 108 |
+
image = torch.as_tensor(np.ascontiguousarray(image.transpose(2, 0, 1)))
|
| 109 |
+
if sem_seg_gt is not None:
|
| 110 |
+
sem_seg_gt = torch.as_tensor(sem_seg_gt.astype("long"))
|
| 111 |
+
pan_seg_gt = torch.as_tensor(pan_seg_gt.astype("long"))
|
| 112 |
+
|
| 113 |
+
if self.size_divisibility > 0:
|
| 114 |
+
image_size = (image.shape[-2], image.shape[-1])
|
| 115 |
+
padding_size = [
|
| 116 |
+
0,
|
| 117 |
+
self.size_divisibility - image_size[1],
|
| 118 |
+
0,
|
| 119 |
+
self.size_divisibility - image_size[0],
|
| 120 |
+
]
|
| 121 |
+
image = F.pad(image, padding_size, value=128).contiguous()
|
| 122 |
+
if sem_seg_gt is not None:
|
| 123 |
+
sem_seg_gt = F.pad(sem_seg_gt, padding_size, value=self.ignore_label).contiguous()
|
| 124 |
+
pan_seg_gt = F.pad(
|
| 125 |
+
pan_seg_gt, padding_size, value=0
|
| 126 |
+
).contiguous() # 0 is the VOID panoptic label
|
| 127 |
+
|
| 128 |
+
image_shape = (image.shape[-2], image.shape[-1]) # h, w
|
| 129 |
+
|
| 130 |
+
# Pytorch's dataloader is efficient on torch.Tensor due to shared-memory,
|
| 131 |
+
# but not efficient on large generic data structures due to the use of pickle & mp.Queue.
|
| 132 |
+
# Therefore it's important to use torch.Tensor.
|
| 133 |
+
dataset_dict["image"] = image
|
| 134 |
+
if sem_seg_gt is not None:
|
| 135 |
+
dataset_dict["sem_seg"] = sem_seg_gt.long()
|
| 136 |
+
|
| 137 |
+
if "annotations" in dataset_dict:
|
| 138 |
+
raise ValueError("Pemantic segmentation dataset should not have 'annotations'.")
|
| 139 |
+
|
| 140 |
+
# Prepare per-category binary masks
|
| 141 |
+
pan_seg_gt = pan_seg_gt.numpy()
|
| 142 |
+
instances = Instances(image_shape)
|
| 143 |
+
classes = []
|
| 144 |
+
masks = []
|
| 145 |
+
for segment_info in segments_info:
|
| 146 |
+
class_id = segment_info["category_id"]
|
| 147 |
+
if not segment_info["iscrowd"]:
|
| 148 |
+
classes.append(class_id)
|
| 149 |
+
masks.append(pan_seg_gt == segment_info["id"])
|
| 150 |
+
|
| 151 |
+
classes = np.array(classes)
|
| 152 |
+
instances.gt_classes = torch.tensor(classes, dtype=torch.int64)
|
| 153 |
+
if len(masks) == 0:
|
| 154 |
+
# Some image does not have annotation (all ignored)
|
| 155 |
+
instances.gt_masks = torch.zeros((0, pan_seg_gt.shape[-2], pan_seg_gt.shape[-1]))
|
| 156 |
+
else:
|
| 157 |
+
masks = BitMasks(
|
| 158 |
+
torch.stack([torch.from_numpy(np.ascontiguousarray(x.copy())) for x in masks])
|
| 159 |
+
)
|
| 160 |
+
instances.gt_masks = masks.tensor
|
| 161 |
+
|
| 162 |
+
dataset_dict["instances"] = instances
|
| 163 |
+
|
| 164 |
+
return dataset_dict
|
frozenseg/data/dataset_mappers/mask_former_semantic_dataset_mapper.py
ADDED
|
@@ -0,0 +1,183 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import copy
|
| 2 |
+
import logging
|
| 3 |
+
|
| 4 |
+
import numpy as np
|
| 5 |
+
import torch
|
| 6 |
+
from torch.nn import functional as F
|
| 7 |
+
|
| 8 |
+
from detectron2.config import configurable
|
| 9 |
+
from detectron2.data import MetadataCatalog
|
| 10 |
+
from detectron2.data import detection_utils as utils
|
| 11 |
+
from detectron2.data import transforms as T
|
| 12 |
+
from detectron2.projects.point_rend import ColorAugSSDTransform
|
| 13 |
+
from detectron2.structures import BitMasks, Instances
|
| 14 |
+
|
| 15 |
+
__all__ = ["MaskFormerSemanticDatasetMapper"]
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
class MaskFormerSemanticDatasetMapper:
|
| 19 |
+
"""
|
| 20 |
+
A callable which takes a dataset dict in Detectron2 Dataset format,
|
| 21 |
+
and map it into a format used by MaskFormer for semantic segmentation.
|
| 22 |
+
|
| 23 |
+
The callable currently does the following:
|
| 24 |
+
|
| 25 |
+
1. Read the image from "file_name"
|
| 26 |
+
2. Applies geometric transforms to the image and annotation
|
| 27 |
+
3. Find and applies suitable cropping to the image and annotation
|
| 28 |
+
4. Prepare image and annotation to Tensors
|
| 29 |
+
"""
|
| 30 |
+
|
| 31 |
+
@configurable
|
| 32 |
+
def __init__(
|
| 33 |
+
self,
|
| 34 |
+
is_train=True,
|
| 35 |
+
*,
|
| 36 |
+
augmentations,
|
| 37 |
+
image_format,
|
| 38 |
+
ignore_label,
|
| 39 |
+
size_divisibility,
|
| 40 |
+
):
|
| 41 |
+
"""
|
| 42 |
+
NOTE: this interface is experimental.
|
| 43 |
+
Args:
|
| 44 |
+
is_train: for training or inference
|
| 45 |
+
augmentations: a list of augmentations or deterministic transforms to apply
|
| 46 |
+
image_format: an image format supported by :func:`detection_utils.read_image`.
|
| 47 |
+
ignore_label: the label that is ignored to evaluation
|
| 48 |
+
size_divisibility: pad image size to be divisible by this value
|
| 49 |
+
"""
|
| 50 |
+
self.is_train = is_train
|
| 51 |
+
self.tfm_gens = augmentations
|
| 52 |
+
self.img_format = image_format
|
| 53 |
+
self.ignore_label = ignore_label
|
| 54 |
+
self.size_divisibility = size_divisibility
|
| 55 |
+
|
| 56 |
+
logger = logging.getLogger(__name__)
|
| 57 |
+
mode = "training" if is_train else "inference"
|
| 58 |
+
logger.info(f"[{self.__class__.__name__}] Augmentations used in {mode}: {augmentations}")
|
| 59 |
+
|
| 60 |
+
@classmethod
|
| 61 |
+
def from_config(cls, cfg, is_train=True):
|
| 62 |
+
# Build augmentation
|
| 63 |
+
augs = [
|
| 64 |
+
T.ResizeShortestEdge(
|
| 65 |
+
cfg.INPUT.MIN_SIZE_TRAIN,
|
| 66 |
+
cfg.INPUT.MAX_SIZE_TRAIN,
|
| 67 |
+
cfg.INPUT.MIN_SIZE_TRAIN_SAMPLING,
|
| 68 |
+
)
|
| 69 |
+
]
|
| 70 |
+
if cfg.INPUT.CROP.ENABLED:
|
| 71 |
+
augs.append(
|
| 72 |
+
T.RandomCrop_CategoryAreaConstraint(
|
| 73 |
+
cfg.INPUT.CROP.TYPE,
|
| 74 |
+
cfg.INPUT.CROP.SIZE,
|
| 75 |
+
cfg.INPUT.CROP.SINGLE_CATEGORY_MAX_AREA,
|
| 76 |
+
cfg.MODEL.SEM_SEG_HEAD.IGNORE_VALUE,
|
| 77 |
+
)
|
| 78 |
+
)
|
| 79 |
+
if cfg.INPUT.COLOR_AUG_SSD:
|
| 80 |
+
augs.append(ColorAugSSDTransform(img_format=cfg.INPUT.FORMAT))
|
| 81 |
+
augs.append(T.RandomFlip())
|
| 82 |
+
|
| 83 |
+
# Assume always applies to the training set.
|
| 84 |
+
dataset_names = cfg.DATASETS.TRAIN
|
| 85 |
+
meta = MetadataCatalog.get(dataset_names[0])
|
| 86 |
+
ignore_label = meta.ignore_label
|
| 87 |
+
|
| 88 |
+
ret = {
|
| 89 |
+
"is_train": is_train,
|
| 90 |
+
"augmentations": augs,
|
| 91 |
+
"image_format": cfg.INPUT.FORMAT,
|
| 92 |
+
"ignore_label": ignore_label,
|
| 93 |
+
"size_divisibility": cfg.INPUT.SIZE_DIVISIBILITY,
|
| 94 |
+
}
|
| 95 |
+
return ret
|
| 96 |
+
|
| 97 |
+
def __call__(self, dataset_dict):
|
| 98 |
+
"""
|
| 99 |
+
Args:
|
| 100 |
+
dataset_dict (dict): Metadata of one image, in Detectron2 Dataset format.
|
| 101 |
+
|
| 102 |
+
Returns:
|
| 103 |
+
dict: a format that builtin models in detectron2 accept
|
| 104 |
+
"""
|
| 105 |
+
assert self.is_train, "MaskFormerSemanticDatasetMapper should only be used for training!"
|
| 106 |
+
|
| 107 |
+
dataset_dict = copy.deepcopy(dataset_dict) # it will be modified by code below
|
| 108 |
+
image = utils.read_image(dataset_dict["file_name"], format=self.img_format)
|
| 109 |
+
utils.check_image_size(dataset_dict, image)
|
| 110 |
+
|
| 111 |
+
if "sem_seg_file_name" in dataset_dict:
|
| 112 |
+
# PyTorch transformation not implemented for uint16, so converting it to double first
|
| 113 |
+
sem_seg_gt = utils.read_image(dataset_dict.pop("sem_seg_file_name")).astype("double")
|
| 114 |
+
else:
|
| 115 |
+
sem_seg_gt = None
|
| 116 |
+
|
| 117 |
+
if sem_seg_gt is None:
|
| 118 |
+
raise ValueError(
|
| 119 |
+
"Cannot find 'sem_seg_file_name' for semantic segmentation dataset {}.".format(
|
| 120 |
+
dataset_dict["file_name"]
|
| 121 |
+
)
|
| 122 |
+
)
|
| 123 |
+
|
| 124 |
+
aug_input = T.AugInput(image, sem_seg=sem_seg_gt)
|
| 125 |
+
aug_input, transforms = T.apply_transform_gens(self.tfm_gens, aug_input)
|
| 126 |
+
image = aug_input.image
|
| 127 |
+
sem_seg_gt = aug_input.sem_seg
|
| 128 |
+
|
| 129 |
+
# Pad image and segmentation label here!
|
| 130 |
+
image = torch.as_tensor(np.ascontiguousarray(image.transpose(2, 0, 1)))
|
| 131 |
+
if sem_seg_gt is not None:
|
| 132 |
+
sem_seg_gt = torch.as_tensor(sem_seg_gt.astype("long"))
|
| 133 |
+
|
| 134 |
+
if self.size_divisibility > 0:
|
| 135 |
+
image_size = (image.shape[-2], image.shape[-1])
|
| 136 |
+
padding_size = [
|
| 137 |
+
0,
|
| 138 |
+
self.size_divisibility - image_size[1],
|
| 139 |
+
0,
|
| 140 |
+
self.size_divisibility - image_size[0],
|
| 141 |
+
]
|
| 142 |
+
image = F.pad(image, padding_size, value=128).contiguous()
|
| 143 |
+
if sem_seg_gt is not None:
|
| 144 |
+
sem_seg_gt = F.pad(sem_seg_gt, padding_size, value=self.ignore_label).contiguous()
|
| 145 |
+
|
| 146 |
+
image_shape = (image.shape[-2], image.shape[-1]) # h, w
|
| 147 |
+
|
| 148 |
+
# Pytorch's dataloader is efficient on torch.Tensor due to shared-memory,
|
| 149 |
+
# but not efficient on large generic data structures due to the use of pickle & mp.Queue.
|
| 150 |
+
# Therefore it's important to use torch.Tensor.
|
| 151 |
+
dataset_dict["image"] = image
|
| 152 |
+
|
| 153 |
+
if sem_seg_gt is not None:
|
| 154 |
+
dataset_dict["sem_seg"] = sem_seg_gt.long()
|
| 155 |
+
|
| 156 |
+
if "annotations" in dataset_dict:
|
| 157 |
+
raise ValueError("Semantic segmentation dataset should not have 'annotations'.")
|
| 158 |
+
|
| 159 |
+
# Prepare per-category binary masks
|
| 160 |
+
if sem_seg_gt is not None:
|
| 161 |
+
sem_seg_gt = sem_seg_gt.numpy()
|
| 162 |
+
instances = Instances(image_shape)
|
| 163 |
+
classes = np.unique(sem_seg_gt)
|
| 164 |
+
# remove ignored region
|
| 165 |
+
classes = classes[classes != self.ignore_label]
|
| 166 |
+
instances.gt_classes = torch.tensor(classes, dtype=torch.int64)
|
| 167 |
+
|
| 168 |
+
masks = []
|
| 169 |
+
for class_id in classes:
|
| 170 |
+
masks.append(sem_seg_gt == class_id)
|
| 171 |
+
|
| 172 |
+
if len(masks) == 0:
|
| 173 |
+
# Some image does not have annotation (all ignored)
|
| 174 |
+
instances.gt_masks = torch.zeros((0, sem_seg_gt.shape[-2], sem_seg_gt.shape[-1]))
|
| 175 |
+
else:
|
| 176 |
+
masks = BitMasks(
|
| 177 |
+
torch.stack([torch.from_numpy(np.ascontiguousarray(x.copy())) for x in masks])
|
| 178 |
+
)
|
| 179 |
+
instances.gt_masks = masks.tensor
|
| 180 |
+
|
| 181 |
+
dataset_dict["instances"] = instances
|
| 182 |
+
|
| 183 |
+
return dataset_dict
|
frozenseg/data/datasets/__init__.py
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from . import (
|
| 2 |
+
register_lvis_instance,
|
| 3 |
+
register_coco_panoptic_annos_semseg,
|
| 4 |
+
register_ade20k_panoptic,
|
| 5 |
+
register_cityscapes_panoptic,
|
| 6 |
+
register_mapillary_vistas_panoptic,
|
| 7 |
+
register_ade20k_full,
|
| 8 |
+
register_pascal_voc_20_semantic,
|
| 9 |
+
register_pascal_voc_21_semantic,
|
| 10 |
+
register_pascal_ctx_59_sem_seg,
|
| 11 |
+
register_pascal_ctx_459_sem_seg,
|
| 12 |
+
register_coco_instance,
|
| 13 |
+
register_ade20k_instance,
|
| 14 |
+
register_coco_stuff_164k,
|
| 15 |
+
openseg_classes,
|
| 16 |
+
register_bdd100k_panoseg,
|
| 17 |
+
register_bdd100k_semseg,
|
| 18 |
+
)
|
frozenseg/data/datasets/ade20k_150_with_prompt_eng.txt
ADDED
|
@@ -0,0 +1,151 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
0:invalid_class_id
|
| 2 |
+
1:wall,walls,brick wall,stone wall,interior wall
|
| 3 |
+
2:building,buildings,edifice,edifices
|
| 4 |
+
3:sky,clouds
|
| 5 |
+
4:floor,flooring
|
| 6 |
+
5:tree,trees
|
| 7 |
+
6:ceiling
|
| 8 |
+
7:road,route,street,roads,streets,routes
|
| 9 |
+
8:bed,beds
|
| 10 |
+
9:windowpane,window,windows
|
| 11 |
+
10:grass,grass field
|
| 12 |
+
11:cabinet,cabinets,wall mounted cabine
|
| 13 |
+
12:sidewalk,pavement
|
| 14 |
+
13:person,child,girl,boy,woman,man,people,children,girls,boys,women,men
|
| 15 |
+
14:earth,ground
|
| 16 |
+
15:door,double door,doors
|
| 17 |
+
16:table,tables,tablecloth
|
| 18 |
+
17:mountain,mount,mountains
|
| 19 |
+
18:plant,flora,plant life,plants,bushes
|
| 20 |
+
19:curtain,drape,drapery,mantle,pall
|
| 21 |
+
20:chair,chairs
|
| 22 |
+
21:car,automobile,cars
|
| 23 |
+
22:water
|
| 24 |
+
23:painting,picture,paintings,pictures,wallart,framed canvas
|
| 25 |
+
24:sofa,couch,sofas,couches
|
| 26 |
+
25:shelf,shelves
|
| 27 |
+
26:house exterior
|
| 28 |
+
27:sea,ocean
|
| 29 |
+
28:mirror,mirrors
|
| 30 |
+
29:rug,carpet,carpeting
|
| 31 |
+
30:field
|
| 32 |
+
31:armchair,armchairs
|
| 33 |
+
32:seat,seats
|
| 34 |
+
33:fence,fencing
|
| 35 |
+
34:desk,desks
|
| 36 |
+
35:rock,stone,rocks,stones
|
| 37 |
+
36:wardrobe,closet,press,wardrobes,closets
|
| 38 |
+
37:lamp,lamps
|
| 39 |
+
38:bathtub,bathing tub,bath,tub
|
| 40 |
+
39:railing,rail
|
| 41 |
+
40:cushion,cushions
|
| 42 |
+
41:pedestal
|
| 43 |
+
42:box,boxes
|
| 44 |
+
43:column,pillar
|
| 45 |
+
44:signboard,sign,signboards,signs
|
| 46 |
+
45:chest of drawers,chest,bureau,dresser
|
| 47 |
+
46:counter
|
| 48 |
+
47:sand
|
| 49 |
+
48:sink
|
| 50 |
+
49:skyscraper,skyscrapers
|
| 51 |
+
50:fireplace,hearth,open fireplace
|
| 52 |
+
51:refrigerator,icebox
|
| 53 |
+
52:grandstand,covered stand
|
| 54 |
+
53:path
|
| 55 |
+
54:stairs,steps
|
| 56 |
+
55:runway
|
| 57 |
+
56:case,display case,showcase,vitrine
|
| 58 |
+
57:pool table,billiard table,snooker table
|
| 59 |
+
58:pillow,pillows
|
| 60 |
+
59:screen door,shower door
|
| 61 |
+
60:stairway,staircase
|
| 62 |
+
61:river
|
| 63 |
+
62:bridge,span
|
| 64 |
+
63:bookcase
|
| 65 |
+
64:window screen,door screen
|
| 66 |
+
65:coffee table,cocktail table
|
| 67 |
+
66:toilet,commode,crapper,potty
|
| 68 |
+
67:flower,flowers
|
| 69 |
+
68:book,books
|
| 70 |
+
69:hill
|
| 71 |
+
70:bench,benches
|
| 72 |
+
71:countertop,counter top,worktop
|
| 73 |
+
72:stove,kitchen stove,kitchen range,kitchen range,cooking stove
|
| 74 |
+
73:palm tree,palm trees
|
| 75 |
+
74:kitchen island
|
| 76 |
+
75:computer,computing machine,computing device,data processor,electronic computer,information processing system
|
| 77 |
+
76:swivel chair
|
| 78 |
+
77:boat
|
| 79 |
+
78:bar
|
| 80 |
+
79:arcade machine,arcade machines
|
| 81 |
+
80:hovel,hut,hutch,shack,shanty
|
| 82 |
+
81:bus,autobus,double-decker,jitney,motorbus,motorcoach,omnibus,passenger vehicle
|
| 83 |
+
82:towel
|
| 84 |
+
83:light bulb,lightbulb,bulb,incandescent lamp,electric light,electric-light bulb
|
| 85 |
+
84:truck,motortruck
|
| 86 |
+
85:tower,towers
|
| 87 |
+
86:chandelier,pendant,pendent
|
| 88 |
+
87:awning,sunshade,sunblind
|
| 89 |
+
88:streetlight,street lamp
|
| 90 |
+
89:booth,cubicle,stall,kiosk
|
| 91 |
+
90:television receiver,television,television set,tv,tv set
|
| 92 |
+
91:airplane,aeroplane,airplanes,aeroplanes
|
| 93 |
+
92:dirt track
|
| 94 |
+
93:apparel,wearing apparel,dress,clothes
|
| 95 |
+
94:pole
|
| 96 |
+
95:land,soil
|
| 97 |
+
96:bannister,banister,balustrade,balusters,handrail
|
| 98 |
+
97:escalator,moving staircase,moving stairway
|
| 99 |
+
98:ottoman,pouf,pouffe,puff,hassock
|
| 100 |
+
99:bottle,bottles,water bottle
|
| 101 |
+
100:buffet,sideboard
|
| 102 |
+
101:poster,posting,placard,notice,bill,card
|
| 103 |
+
102:stage
|
| 104 |
+
103:van
|
| 105 |
+
104:ship
|
| 106 |
+
105:fountain
|
| 107 |
+
106:conveyer belt,conveyor belt,conveyer,conveyor,transporter
|
| 108 |
+
107:canopy
|
| 109 |
+
108:washer,automatic washer,washing machine
|
| 110 |
+
109:plaything,toy,toys
|
| 111 |
+
110:swimming pool,swimming bath
|
| 112 |
+
111:stool,stools
|
| 113 |
+
112:barrel,cask,barrels,casks
|
| 114 |
+
113:basket,handbasket
|
| 115 |
+
114:waterfall,falls
|
| 116 |
+
115:tent,collapsible shelter
|
| 117 |
+
116:bag,bags,gift bag,paper bag
|
| 118 |
+
117:minibike,motorbike
|
| 119 |
+
118:cradle
|
| 120 |
+
119:oven
|
| 121 |
+
120:ball,balls
|
| 122 |
+
121:food,solid food
|
| 123 |
+
122:step,stair
|
| 124 |
+
123:tank,storage tank
|
| 125 |
+
124:trade name,brand name,brand,marque
|
| 126 |
+
125:microwave,microwave oven
|
| 127 |
+
126:plant pots,plant pot,flower pot,flowerpot,planter
|
| 128 |
+
127:animal,animate being,dog,cat,horse,cow,sheep,zebra,girraffe,bird
|
| 129 |
+
128:bicycle,bike
|
| 130 |
+
129:lake
|
| 131 |
+
130:dishwasher,dish washer,dishwashing machine
|
| 132 |
+
131:projection screen
|
| 133 |
+
132:blanket,cover
|
| 134 |
+
133:sculpture,sculptures
|
| 135 |
+
134:exhaust hood
|
| 136 |
+
135:sconce,sconce lamp,sconce light
|
| 137 |
+
136:vase,vases
|
| 138 |
+
137:traffic light,traffic signal,traffic lights
|
| 139 |
+
138:tray,trays
|
| 140 |
+
139:ashcan,trash can,garbage can,wastebin,ash bin,ash-bin,ashbin,dustbin,trash barrel,trash bin
|
| 141 |
+
140:ceiling fan,floor fan
|
| 142 |
+
141:pier,wharf,wharfage,dock
|
| 143 |
+
142:crt screen
|
| 144 |
+
143:plate,plates
|
| 145 |
+
144:monitor,monitoring device,monitors
|
| 146 |
+
145:bulletin board,notice board
|
| 147 |
+
146:shower
|
| 148 |
+
147:radiator
|
| 149 |
+
148:cup,cups,drinking glass,drinking glasses
|
| 150 |
+
149:clock
|
| 151 |
+
150:flag,flags
|
frozenseg/data/datasets/ade20k_847_with_prompt_eng.txt
ADDED
|
@@ -0,0 +1,848 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
0:invalid_class_id
|
| 2 |
+
1:wall,walls,interior wall,brick wall,stone wall
|
| 3 |
+
2:building,buildings,edifice,edifices
|
| 4 |
+
3:sky,clouds
|
| 5 |
+
4:tree,trees
|
| 6 |
+
5:road,route,street,roads,streets,routes
|
| 7 |
+
6:floor,flooring
|
| 8 |
+
7:ceiling
|
| 9 |
+
8:bed,beds
|
| 10 |
+
9:sidewalk,pavement
|
| 11 |
+
10:earth,ground
|
| 12 |
+
11:cabinet,cabinets,wall mounted cabine
|
| 13 |
+
12:person,child,girl,boy,woman,man,people,children,girls,boys,women,men
|
| 14 |
+
13:grass,grass field
|
| 15 |
+
14:windowpane,window,windows
|
| 16 |
+
15:car,automobile,cars
|
| 17 |
+
16:mountain,mount,mountains
|
| 18 |
+
17:plant,flora,plant life,plants,bushes
|
| 19 |
+
18:table,tables,tablecloth
|
| 20 |
+
19:chair,chairs
|
| 21 |
+
20:curtain,drape,drapery,mantle,pall
|
| 22 |
+
21:door,double door,doors
|
| 23 |
+
22:sofa,couch,sofas,couches
|
| 24 |
+
23:sea,ocean
|
| 25 |
+
24:painting,picture,paintings,pictures,wallart,framed canvas
|
| 26 |
+
25:water
|
| 27 |
+
26:mirror,mirrors
|
| 28 |
+
27:house exterior
|
| 29 |
+
28:rug,carpet,carpeting
|
| 30 |
+
29:shelf,shelves
|
| 31 |
+
30:armchair,armchairs
|
| 32 |
+
31:fence,fencing
|
| 33 |
+
32:field
|
| 34 |
+
33:lamp,lamps
|
| 35 |
+
34:rock,stone,rocks,stones
|
| 36 |
+
35:seat,seats
|
| 37 |
+
36:river
|
| 38 |
+
37:desk,desks
|
| 39 |
+
38:bathtub,bathing tub,bath,tub
|
| 40 |
+
39:railing,rail
|
| 41 |
+
40:signboard,sign,signboards,signs
|
| 42 |
+
41:cushion,cushions
|
| 43 |
+
42:path
|
| 44 |
+
43:work surface
|
| 45 |
+
44:stairs,steps
|
| 46 |
+
45:column,pillar
|
| 47 |
+
46:sink
|
| 48 |
+
47:wardrobe,closet,press,wardrobes,closets
|
| 49 |
+
48:snow
|
| 50 |
+
49:refrigerator,icebox
|
| 51 |
+
50:pedestal
|
| 52 |
+
51:bridge,span
|
| 53 |
+
52:blind
|
| 54 |
+
53:runway
|
| 55 |
+
54:cliff,drop,drop-off
|
| 56 |
+
55:sand
|
| 57 |
+
56:fireplace,hearth,open fireplace
|
| 58 |
+
57:pillow,pillows
|
| 59 |
+
58:screen door,shower door
|
| 60 |
+
59:toilet,commode,crapper,potty
|
| 61 |
+
60:skyscraper,skyscrapers
|
| 62 |
+
61:grandstand,covered stand
|
| 63 |
+
62:box,boxes
|
| 64 |
+
63:pool table,billiard table,snooker table
|
| 65 |
+
64:palm tree,palm trees
|
| 66 |
+
65:double door
|
| 67 |
+
66:coffee table,cocktail table
|
| 68 |
+
67:counter
|
| 69 |
+
68:countertop,counter top,worktop
|
| 70 |
+
69:chest of drawers,chest,bureau,dresser
|
| 71 |
+
70:kitchen island
|
| 72 |
+
71:boat
|
| 73 |
+
72:waterfall,falls
|
| 74 |
+
73:stove,kitchen stove,kitchen range,kitchen range,cooking stove
|
| 75 |
+
74:flower,flowers
|
| 76 |
+
75:bookcase
|
| 77 |
+
76:controls
|
| 78 |
+
77:book,books
|
| 79 |
+
78:stairway,staircase
|
| 80 |
+
79:streetlight,street lamp
|
| 81 |
+
80:computer,computing machine,computing device,data processor,electronic computer,information processing system
|
| 82 |
+
81:bus,autobus,double-decker,jitney,motorbus,motorcoach,omnibus,passenger vehicle
|
| 83 |
+
82:swivel chair
|
| 84 |
+
83:light,light source
|
| 85 |
+
84:bench,benches
|
| 86 |
+
85:case,display case,showcase,vitrine
|
| 87 |
+
86:towel
|
| 88 |
+
87:fountain
|
| 89 |
+
88:embankment
|
| 90 |
+
89:television receiver,television,television set,tv,tv set
|
| 91 |
+
90:van
|
| 92 |
+
91:hill
|
| 93 |
+
92:awning,sunshade,sunblind
|
| 94 |
+
93:poster,posting,placard,notice,bill,card
|
| 95 |
+
94:truck,motortruck
|
| 96 |
+
95:airplane,aeroplane,airplanes,aeroplanes
|
| 97 |
+
96:pole
|
| 98 |
+
97:tower,towers
|
| 99 |
+
98:court
|
| 100 |
+
99:ball,balls
|
| 101 |
+
100:aircraft carrier,carrier,flattop,attack aircraft carrier
|
| 102 |
+
101:buffet,sideboard
|
| 103 |
+
102:hovel,hut,hutch,shack,shanty
|
| 104 |
+
103:apparel,wearing apparel,dress,clothes
|
| 105 |
+
104:minibike,motorbike
|
| 106 |
+
105:animal,animate being,dog,cat,horse,cow,sheep,zebra,giraffe,bird
|
| 107 |
+
106:chandelier,pendant,pendent
|
| 108 |
+
107:step,stair
|
| 109 |
+
108:booth,cubicle,stall,kiosk
|
| 110 |
+
109:bicycle,bike
|
| 111 |
+
110:doorframe,doorcase
|
| 112 |
+
111:sconce,sconce lamp,sconce light
|
| 113 |
+
112:pond
|
| 114 |
+
113:trade name,brand name
|
| 115 |
+
114:bannister,banister,balustrade,balusters,handrail
|
| 116 |
+
115:bag,bags,gift bag,paper bag
|
| 117 |
+
116:traffic light,traffic signal,traffic lights
|
| 118 |
+
117:gazebo
|
| 119 |
+
118:escalator,moving staircase,moving stairway
|
| 120 |
+
119:land,soil
|
| 121 |
+
120:board,plank
|
| 122 |
+
121:arcade machine,arcade machines
|
| 123 |
+
122:eiderdown,duvet,continental quilt
|
| 124 |
+
123:bar
|
| 125 |
+
124:stall,stand,sales booth
|
| 126 |
+
125:playground
|
| 127 |
+
126:ship
|
| 128 |
+
127:ottoman,pouf,pouffe,puff,hassock
|
| 129 |
+
128:ashcan,trash can,garbage can,wastebin,ash bin,ash-bin,ashbin,dustbin,trash barrel,trash bin
|
| 130 |
+
129:bottle,bottles,water bottle
|
| 131 |
+
130:cradle
|
| 132 |
+
131:pot,flowerpot
|
| 133 |
+
132:conveyer belt,conveyor belt,conveyer,conveyor,transporter
|
| 134 |
+
133:train,railroad train
|
| 135 |
+
134:stool,stools
|
| 136 |
+
135:lake
|
| 137 |
+
136:tank,storage tank
|
| 138 |
+
137:ice,water ice
|
| 139 |
+
138:basket,handbasket
|
| 140 |
+
139:manhole
|
| 141 |
+
140:tent,collapsible shelter
|
| 142 |
+
141:canopy
|
| 143 |
+
142:microwave,microwave oven
|
| 144 |
+
143:barrel,cask,barrels,casks
|
| 145 |
+
144:dirt track
|
| 146 |
+
145:beam
|
| 147 |
+
146:dishwasher,dish washer,dishwashing machine
|
| 148 |
+
147:plate,plates
|
| 149 |
+
148:crt screen
|
| 150 |
+
149:ruins
|
| 151 |
+
150:washer,automatic washer,washing machine
|
| 152 |
+
151:blanket,cover
|
| 153 |
+
152:plaything,toy,toys
|
| 154 |
+
153:food,solid food
|
| 155 |
+
154:projection screen
|
| 156 |
+
155:oven
|
| 157 |
+
156:stage
|
| 158 |
+
157:beacon,lighthouse,beacon light,pharos
|
| 159 |
+
158:umbrella
|
| 160 |
+
159:sculpture,sculptures
|
| 161 |
+
160:aqueduct
|
| 162 |
+
161:container
|
| 163 |
+
162:scaffolding,staging
|
| 164 |
+
163:exhaust hood
|
| 165 |
+
164:curb,curbing,kerb
|
| 166 |
+
165:roller coaster
|
| 167 |
+
166:horse,equus caballus
|
| 168 |
+
167:catwalk
|
| 169 |
+
168:glass,drinking glass
|
| 170 |
+
169:vase,vases
|
| 171 |
+
170:central reservation
|
| 172 |
+
171:carousel
|
| 173 |
+
172:radiator
|
| 174 |
+
173:closet
|
| 175 |
+
174:machine
|
| 176 |
+
175:pier,wharf,wharfage,dock
|
| 177 |
+
176:ceiling fan,floor fan
|
| 178 |
+
177:inflatable bounce game
|
| 179 |
+
178:pitch
|
| 180 |
+
179:paper
|
| 181 |
+
180:arcade,colonnade
|
| 182 |
+
181:hot tub
|
| 183 |
+
182:helicopter
|
| 184 |
+
183:tray,trays
|
| 185 |
+
184:partition,divider
|
| 186 |
+
185:vineyard
|
| 187 |
+
186:bowl
|
| 188 |
+
187:bullring
|
| 189 |
+
188:flag,flags
|
| 190 |
+
189:pot
|
| 191 |
+
190:footbridge,overcrossing,pedestrian bridge
|
| 192 |
+
191:shower
|
| 193 |
+
192:bag,traveling bag,travelling bag,grip,suitcase
|
| 194 |
+
193:bulletin board,notice board
|
| 195 |
+
194:confessional booth
|
| 196 |
+
195:trunk,tree trunk,bole
|
| 197 |
+
196:forest
|
| 198 |
+
197:elevator door
|
| 199 |
+
198:laptop,laptop computer
|
| 200 |
+
199:instrument panel
|
| 201 |
+
200:bucket,pail
|
| 202 |
+
201:tapestry,tapis
|
| 203 |
+
202:platform
|
| 204 |
+
203:jacket
|
| 205 |
+
204:gate
|
| 206 |
+
205:monitor,monitoring device,monitors
|
| 207 |
+
206:telephone booth,phone booth,call box,telephone box,telephone kiosk
|
| 208 |
+
207:spotlight,spot
|
| 209 |
+
208:ring
|
| 210 |
+
209:control panel
|
| 211 |
+
210:blackboard,chalkboard
|
| 212 |
+
211:air conditioner,air conditioning
|
| 213 |
+
212:chest
|
| 214 |
+
213:clock
|
| 215 |
+
214:sand dune
|
| 216 |
+
215:pipe,pipage,piping
|
| 217 |
+
216:vault
|
| 218 |
+
217:table football
|
| 219 |
+
218:cannon
|
| 220 |
+
219:swimming pool,swimming bath
|
| 221 |
+
220:fluorescent,fluorescent fixture
|
| 222 |
+
221:statue
|
| 223 |
+
222:loudspeaker,speaker,speaker unit,loudspeaker system,speaker system
|
| 224 |
+
223:exhibitor
|
| 225 |
+
224:ladder
|
| 226 |
+
225:carport
|
| 227 |
+
226:dam
|
| 228 |
+
227:pulpit
|
| 229 |
+
228:skylight,fanlight
|
| 230 |
+
229:water tower
|
| 231 |
+
230:grill,grille,grillwork
|
| 232 |
+
231:display board
|
| 233 |
+
232:pane,pane of glass,window glass
|
| 234 |
+
233:rubbish,trash,scrap
|
| 235 |
+
234:ice rink
|
| 236 |
+
235:fruit
|
| 237 |
+
236:patio
|
| 238 |
+
237:vending machine
|
| 239 |
+
238:telephone,phone,telephone set
|
| 240 |
+
239:net
|
| 241 |
+
240:backpack,back pack,knapsack,packsack,rucksack,haversack
|
| 242 |
+
241:jar
|
| 243 |
+
242:track
|
| 244 |
+
243:magazine
|
| 245 |
+
244:shutter
|
| 246 |
+
245:roof
|
| 247 |
+
246:banner,streamer
|
| 248 |
+
247:landfill
|
| 249 |
+
248:post
|
| 250 |
+
249:altarpiece,reredos
|
| 251 |
+
250:hat,chapeau,lid
|
| 252 |
+
251:arch,archway
|
| 253 |
+
252:table game
|
| 254 |
+
253:bag,handbag,pocketbook,purse
|
| 255 |
+
254:document,written document,papers
|
| 256 |
+
255:dome
|
| 257 |
+
256:pier
|
| 258 |
+
257:shanties
|
| 259 |
+
258:forecourt
|
| 260 |
+
259:crane
|
| 261 |
+
260:dog,domestic dog,canis familiaris
|
| 262 |
+
261:piano,pianoforte,forte-piano
|
| 263 |
+
262:drawing
|
| 264 |
+
263:cabin
|
| 265 |
+
264:ad,advertisement,advertizement,advertising,advertizing,advert
|
| 266 |
+
265:amphitheater,amphitheatre,coliseum
|
| 267 |
+
266:monument
|
| 268 |
+
267:henhouse
|
| 269 |
+
268:cockpit
|
| 270 |
+
269:heater,warmer
|
| 271 |
+
270:windmill,aerogenerator,wind generator
|
| 272 |
+
271:pool
|
| 273 |
+
272:elevator,lift
|
| 274 |
+
273:decoration,ornament,ornamentation
|
| 275 |
+
274:labyrinth
|
| 276 |
+
275:text,textual matter
|
| 277 |
+
276:printer
|
| 278 |
+
277:mezzanine,first balcony
|
| 279 |
+
278:mattress
|
| 280 |
+
279:straw
|
| 281 |
+
280:stalls
|
| 282 |
+
281:patio,terrace
|
| 283 |
+
282:billboard,hoarding
|
| 284 |
+
283:bus stop
|
| 285 |
+
284:trouser,pant
|
| 286 |
+
285:console table,console
|
| 287 |
+
286:rack
|
| 288 |
+
287:notebook
|
| 289 |
+
288:shrine
|
| 290 |
+
289:pantry
|
| 291 |
+
290:cart
|
| 292 |
+
291:steam shovel
|
| 293 |
+
292:porch
|
| 294 |
+
293:postbox,mailbox,letter box
|
| 295 |
+
294:figurine,statuette
|
| 296 |
+
295:recycling bin
|
| 297 |
+
296:folding screen
|
| 298 |
+
297:telescope
|
| 299 |
+
298:deck chair,beach chair
|
| 300 |
+
299:kennel
|
| 301 |
+
300:coffee maker
|
| 302 |
+
301:altar,communion table,lord's table
|
| 303 |
+
302:fish
|
| 304 |
+
303:easel
|
| 305 |
+
304:artificial golf green
|
| 306 |
+
305:iceberg
|
| 307 |
+
306:candlestick,candle holder
|
| 308 |
+
307:shower stall,shower bath
|
| 309 |
+
308:television stand
|
| 310 |
+
309:wall socket,wall plug,electric outlet,electrical outlet,outlet,electric receptacle
|
| 311 |
+
310:skeleton
|
| 312 |
+
311:grand piano,grand
|
| 313 |
+
312:candy,confect
|
| 314 |
+
313:grille door
|
| 315 |
+
314:pedestal,plinth,footstall
|
| 316 |
+
315:jersey,t-shirt,tee shirt
|
| 317 |
+
316:shoe
|
| 318 |
+
317:gravestone,headstone,tombstone
|
| 319 |
+
318:shanty
|
| 320 |
+
319:structure
|
| 321 |
+
320:rocking chair,rocker
|
| 322 |
+
321:bird
|
| 323 |
+
322:place mat
|
| 324 |
+
323:tomb
|
| 325 |
+
324:big top
|
| 326 |
+
325:gas pump,gasoline pump,petrol pump,island dispenser
|
| 327 |
+
326:lockers
|
| 328 |
+
327:cage
|
| 329 |
+
328:finger
|
| 330 |
+
329:bleachers
|
| 331 |
+
330:ferris wheel
|
| 332 |
+
331:hairdresser chair
|
| 333 |
+
332:mat
|
| 334 |
+
333:stands
|
| 335 |
+
334:aquarium,fish tank,marine museum
|
| 336 |
+
335:streetcar,tram,tramcar,trolley,trolley car
|
| 337 |
+
336:napkin,table napkin,serviette
|
| 338 |
+
337:dummy
|
| 339 |
+
338:booklet,brochure,folder,leaflet,pamphlet
|
| 340 |
+
339:sand trap
|
| 341 |
+
340:shop,store
|
| 342 |
+
341:table cloth
|
| 343 |
+
342:service station
|
| 344 |
+
343:coffin
|
| 345 |
+
344:drawer
|
| 346 |
+
345:cages
|
| 347 |
+
346:slot machine,coin machine
|
| 348 |
+
347:balcony
|
| 349 |
+
348:volleyball court
|
| 350 |
+
349:table tennis
|
| 351 |
+
350:control table
|
| 352 |
+
351:shirt
|
| 353 |
+
352:merchandise,ware,product
|
| 354 |
+
353:railway
|
| 355 |
+
354:parterre
|
| 356 |
+
355:chimney
|
| 357 |
+
356:can,tin,tin can
|
| 358 |
+
357:tanks
|
| 359 |
+
358:fabric,cloth,material,textile
|
| 360 |
+
359:alga,algae
|
| 361 |
+
360:system
|
| 362 |
+
361:map
|
| 363 |
+
362:greenhouse
|
| 364 |
+
363:mug
|
| 365 |
+
364:barbecue
|
| 366 |
+
365:trailer
|
| 367 |
+
366:toilet tissue,toilet paper,bathroom tissue
|
| 368 |
+
367:organ
|
| 369 |
+
368:dishrag,dishcloth
|
| 370 |
+
369:island
|
| 371 |
+
370:keyboard
|
| 372 |
+
371:trench
|
| 373 |
+
372:basket,basketball hoop,hoop
|
| 374 |
+
373:steering wheel,wheel
|
| 375 |
+
374:pitcher,ewer
|
| 376 |
+
375:goal
|
| 377 |
+
376:bread,breadstuff,staff of life
|
| 378 |
+
377:beds
|
| 379 |
+
378:wood
|
| 380 |
+
379:file cabinet
|
| 381 |
+
380:newspaper,paper
|
| 382 |
+
381:motorboat
|
| 383 |
+
382:rope
|
| 384 |
+
383:guitar
|
| 385 |
+
384:rubble
|
| 386 |
+
385:scarf
|
| 387 |
+
386:barrels
|
| 388 |
+
387:cap
|
| 389 |
+
388:leaves
|
| 390 |
+
389:control tower
|
| 391 |
+
390:dashboard
|
| 392 |
+
391:bandstand
|
| 393 |
+
392:lectern
|
| 394 |
+
393:switch,electric switch,electrical switch
|
| 395 |
+
394:baseboard,mopboard,skirting board
|
| 396 |
+
395:shower room
|
| 397 |
+
396:smoke
|
| 398 |
+
397:faucet,spigot
|
| 399 |
+
398:bulldozer
|
| 400 |
+
399:saucepan
|
| 401 |
+
400:shops
|
| 402 |
+
401:meter
|
| 403 |
+
402:crevasse
|
| 404 |
+
403:gear
|
| 405 |
+
404:candelabrum,candelabra
|
| 406 |
+
405:sofa bed
|
| 407 |
+
406:tunnel
|
| 408 |
+
407:pallet
|
| 409 |
+
408:wire,conducting wire
|
| 410 |
+
409:kettle,boiler
|
| 411 |
+
410:bidet
|
| 412 |
+
411:baby buggy,baby carriage,carriage,perambulator,pram,stroller,go-cart,pushchair,pusher
|
| 413 |
+
412:music stand
|
| 414 |
+
413:pipe,tube
|
| 415 |
+
414:cup,cups,drinking glass,drinking glasses
|
| 416 |
+
415:parking meter
|
| 417 |
+
416:ice hockey rink
|
| 418 |
+
417:shelter
|
| 419 |
+
418:weeds
|
| 420 |
+
419:temple
|
| 421 |
+
420:patty,cake
|
| 422 |
+
421:ski slope
|
| 423 |
+
422:panel
|
| 424 |
+
423:wallet
|
| 425 |
+
424:wheel
|
| 426 |
+
425:towel rack,towel horse
|
| 427 |
+
426:roundabout
|
| 428 |
+
427:canister,cannister,tin
|
| 429 |
+
428:rod
|
| 430 |
+
429:soap dispenser
|
| 431 |
+
430:bell
|
| 432 |
+
431:canvas
|
| 433 |
+
432:box office,ticket office,ticket booth
|
| 434 |
+
433:teacup
|
| 435 |
+
434:trellis
|
| 436 |
+
435:workbench
|
| 437 |
+
436:valley,vale
|
| 438 |
+
437:toaster
|
| 439 |
+
438:knife
|
| 440 |
+
439:podium
|
| 441 |
+
440:ramp
|
| 442 |
+
441:tumble dryer
|
| 443 |
+
442:fireplug,fire hydrant,plug
|
| 444 |
+
443:gym shoe,sneaker,tennis shoe
|
| 445 |
+
444:lab bench
|
| 446 |
+
445:equipment
|
| 447 |
+
446:rocky formation
|
| 448 |
+
447:plastic
|
| 449 |
+
448:calendar
|
| 450 |
+
449:caravan
|
| 451 |
+
450:check-in-desk
|
| 452 |
+
451:ticket counter
|
| 453 |
+
452:brush
|
| 454 |
+
453:mill
|
| 455 |
+
454:covered bridge
|
| 456 |
+
455:bowling alley
|
| 457 |
+
456:hanger
|
| 458 |
+
457:excavator
|
| 459 |
+
458:trestle
|
| 460 |
+
459:revolving door
|
| 461 |
+
460:blast furnace
|
| 462 |
+
461:scale,weighing machine
|
| 463 |
+
462:projector
|
| 464 |
+
463:soap
|
| 465 |
+
464:locker
|
| 466 |
+
465:tractor
|
| 467 |
+
466:stretcher
|
| 468 |
+
467:frame
|
| 469 |
+
468:grating
|
| 470 |
+
469:alembic
|
| 471 |
+
470:candle,taper,wax light
|
| 472 |
+
471:barrier
|
| 473 |
+
472:cardboard
|
| 474 |
+
473:cave
|
| 475 |
+
474:puddle
|
| 476 |
+
475:tarp
|
| 477 |
+
476:price tag
|
| 478 |
+
477:watchtower
|
| 479 |
+
478:meters
|
| 480 |
+
479:light bulb,bulb,bulbs
|
| 481 |
+
480:tracks
|
| 482 |
+
481:hair dryer
|
| 483 |
+
482:skirt
|
| 484 |
+
483:viaduct
|
| 485 |
+
484:paper towel
|
| 486 |
+
485:coat
|
| 487 |
+
486:sheet
|
| 488 |
+
487:fire extinguisher,extinguisher,asphyxiator
|
| 489 |
+
488:water wheel
|
| 490 |
+
489:pottery,clayware
|
| 491 |
+
490:magazine rack
|
| 492 |
+
491:teapot
|
| 493 |
+
492:microphone,mike
|
| 494 |
+
493:support
|
| 495 |
+
494:forklift
|
| 496 |
+
495:canyon
|
| 497 |
+
496:cash register,register
|
| 498 |
+
497:leaf,leafage,foliage
|
| 499 |
+
498:remote control,remote
|
| 500 |
+
499:soap dish
|
| 501 |
+
500:windshield,windscreen
|
| 502 |
+
501:cat
|
| 503 |
+
502:cue,cue stick,pool cue,pool stick
|
| 504 |
+
503:vent,venthole,vent-hole,blowhole
|
| 505 |
+
504:videos
|
| 506 |
+
505:shovel
|
| 507 |
+
506:eaves
|
| 508 |
+
507:antenna,aerial,transmitting aerial
|
| 509 |
+
508:shipyard
|
| 510 |
+
509:hen,biddy
|
| 511 |
+
510:traffic cone
|
| 512 |
+
511:washing machines
|
| 513 |
+
512:truck crane
|
| 514 |
+
513:cds
|
| 515 |
+
514:niche
|
| 516 |
+
515:scoreboard
|
| 517 |
+
516:briefcase
|
| 518 |
+
517:boot
|
| 519 |
+
518:sweater,jumper
|
| 520 |
+
519:hay
|
| 521 |
+
520:pack
|
| 522 |
+
521:bottle rack
|
| 523 |
+
522:glacier
|
| 524 |
+
523:pergola
|
| 525 |
+
524:building materials
|
| 526 |
+
525:television camera
|
| 527 |
+
526:first floor
|
| 528 |
+
527:rifle
|
| 529 |
+
528:tennis table
|
| 530 |
+
529:stadium
|
| 531 |
+
530:safety belt
|
| 532 |
+
531:cover
|
| 533 |
+
532:dish rack
|
| 534 |
+
533:synthesizer
|
| 535 |
+
534:pumpkin
|
| 536 |
+
535:gutter
|
| 537 |
+
536:fruit stand
|
| 538 |
+
537:ice floe,floe
|
| 539 |
+
538:handle,grip,handgrip,hold
|
| 540 |
+
539:wheelchair
|
| 541 |
+
540:mousepad,mouse mat
|
| 542 |
+
541:diploma
|
| 543 |
+
542:fairground ride
|
| 544 |
+
543:radio
|
| 545 |
+
544:hotplate
|
| 546 |
+
545:junk
|
| 547 |
+
546:wheelbarrow
|
| 548 |
+
547:stream
|
| 549 |
+
548:toll plaza
|
| 550 |
+
549:punching bag
|
| 551 |
+
550:trough
|
| 552 |
+
551:throne
|
| 553 |
+
552:chair desk
|
| 554 |
+
553:weighbridge
|
| 555 |
+
554:extractor fan
|
| 556 |
+
555:hanging clothes
|
| 557 |
+
556:dish,dish aerial,dish antenna,saucer
|
| 558 |
+
557:alarm clock,alarm
|
| 559 |
+
558:ski lift
|
| 560 |
+
559:chain
|
| 561 |
+
560:garage
|
| 562 |
+
561:mechanical shovel
|
| 563 |
+
562:wine rack
|
| 564 |
+
563:tramway
|
| 565 |
+
564:treadmill
|
| 566 |
+
565:menu
|
| 567 |
+
566:block
|
| 568 |
+
567:well
|
| 569 |
+
568:witness stand
|
| 570 |
+
569:branch
|
| 571 |
+
570:duck
|
| 572 |
+
571:casserole
|
| 573 |
+
572:frying pan
|
| 574 |
+
573:desk organizer
|
| 575 |
+
574:mast
|
| 576 |
+
575:spectacles,specs,eyeglasses,glasses
|
| 577 |
+
576:service elevator
|
| 578 |
+
577:dollhouse
|
| 579 |
+
578:hammock
|
| 580 |
+
579:clothes hanging
|
| 581 |
+
580:photocopier
|
| 582 |
+
581:notepad
|
| 583 |
+
582:golf cart
|
| 584 |
+
583:footpath
|
| 585 |
+
584:cross
|
| 586 |
+
585:baptismal font
|
| 587 |
+
586:boiler
|
| 588 |
+
587:skip
|
| 589 |
+
588:rotisserie
|
| 590 |
+
589:tables
|
| 591 |
+
590:water mill
|
| 592 |
+
591:helmet
|
| 593 |
+
592:cover curtain
|
| 594 |
+
593:brick
|
| 595 |
+
594:table runner
|
| 596 |
+
595:ashtray
|
| 597 |
+
596:street box
|
| 598 |
+
597:stick
|
| 599 |
+
598:hangers
|
| 600 |
+
599:cells
|
| 601 |
+
600:urinal
|
| 602 |
+
601:centerpiece
|
| 603 |
+
602:portable fridge
|
| 604 |
+
603:dvds
|
| 605 |
+
604:golf club
|
| 606 |
+
605:skirting board
|
| 607 |
+
606:water cooler
|
| 608 |
+
607:clipboard
|
| 609 |
+
608:camera,photographic camera
|
| 610 |
+
609:pigeonhole
|
| 611 |
+
610:chips
|
| 612 |
+
611:food processor
|
| 613 |
+
612:post box
|
| 614 |
+
613:lid
|
| 615 |
+
614:drum
|
| 616 |
+
615:blender
|
| 617 |
+
616:cave entrance
|
| 618 |
+
617:dental chair
|
| 619 |
+
618:obelisk
|
| 620 |
+
619:canoe
|
| 621 |
+
620:mobile
|
| 622 |
+
621:monitors
|
| 623 |
+
622:pool ball
|
| 624 |
+
623:cue rack
|
| 625 |
+
624:baggage carts
|
| 626 |
+
625:shore
|
| 627 |
+
626:fork
|
| 628 |
+
627:paper filer
|
| 629 |
+
628:bicycle rack
|
| 630 |
+
629:coat rack
|
| 631 |
+
630:garland
|
| 632 |
+
631:sports bag
|
| 633 |
+
632:fish tank
|
| 634 |
+
633:towel dispenser
|
| 635 |
+
634:carriage
|
| 636 |
+
635:brochure
|
| 637 |
+
636:plaque
|
| 638 |
+
637:stringer
|
| 639 |
+
638:iron
|
| 640 |
+
639:spoon
|
| 641 |
+
640:flag pole
|
| 642 |
+
641:toilet brush
|
| 643 |
+
642:book stand
|
| 644 |
+
643:water faucet,water tap,tap,hydrant
|
| 645 |
+
644:ticket office
|
| 646 |
+
645:broom
|
| 647 |
+
646:dvd
|
| 648 |
+
647:ice bucket
|
| 649 |
+
648:carapace,shell,cuticle,shield
|
| 650 |
+
649:tureen
|
| 651 |
+
650:folders
|
| 652 |
+
651:chess
|
| 653 |
+
652:root
|
| 654 |
+
653:sewing machine
|
| 655 |
+
654:model
|
| 656 |
+
655:pen
|
| 657 |
+
656:violin
|
| 658 |
+
657:sweatshirt
|
| 659 |
+
658:recycling materials
|
| 660 |
+
659:mitten
|
| 661 |
+
660:chopping board,cutting board
|
| 662 |
+
661:mask
|
| 663 |
+
662:log
|
| 664 |
+
663:mouse,computer mouse
|
| 665 |
+
664:grill
|
| 666 |
+
665:hole
|
| 667 |
+
666:target
|
| 668 |
+
667:trash bag
|
| 669 |
+
668:chalk
|
| 670 |
+
669:sticks
|
| 671 |
+
670:balloon
|
| 672 |
+
671:score
|
| 673 |
+
672:hair spray
|
| 674 |
+
673:roll
|
| 675 |
+
674:runner
|
| 676 |
+
675:engine
|
| 677 |
+
676:inflatable glove
|
| 678 |
+
677:games
|
| 679 |
+
678:pallets
|
| 680 |
+
679:baskets
|
| 681 |
+
680:coop
|
| 682 |
+
681:dvd player
|
| 683 |
+
682:rocking horse
|
| 684 |
+
683:buckets
|
| 685 |
+
684:bread rolls
|
| 686 |
+
685:shawl
|
| 687 |
+
686:watering can
|
| 688 |
+
687:spotlights
|
| 689 |
+
688:post-it
|
| 690 |
+
689:bowls
|
| 691 |
+
690:security camera
|
| 692 |
+
691:runner cloth
|
| 693 |
+
692:lock
|
| 694 |
+
693:alarm,warning device,alarm system
|
| 695 |
+
694:side
|
| 696 |
+
695:roulette
|
| 697 |
+
696:bone
|
| 698 |
+
697:cutlery
|
| 699 |
+
698:pool balls
|
| 700 |
+
699:wheels
|
| 701 |
+
700:spice rack
|
| 702 |
+
701:plant pots,plant pot,flower pot,flowerpot,planter
|
| 703 |
+
702:towel ring
|
| 704 |
+
703:bread box
|
| 705 |
+
704:video
|
| 706 |
+
705:funfair
|
| 707 |
+
706:breads
|
| 708 |
+
707:tripod
|
| 709 |
+
708:ironing board
|
| 710 |
+
709:skimmer
|
| 711 |
+
710:hollow
|
| 712 |
+
711:scratching post
|
| 713 |
+
712:tricycle
|
| 714 |
+
713:file box
|
| 715 |
+
714:mountain pass
|
| 716 |
+
715:tombstones
|
| 717 |
+
716:cooker
|
| 718 |
+
717:card game,cards
|
| 719 |
+
718:golf bag
|
| 720 |
+
719:towel paper
|
| 721 |
+
720:chaise lounge
|
| 722 |
+
721:sun
|
| 723 |
+
722:toilet paper holder
|
| 724 |
+
723:rake
|
| 725 |
+
724:key
|
| 726 |
+
725:umbrella stand
|
| 727 |
+
726:dartboard
|
| 728 |
+
727:transformer
|
| 729 |
+
728:fireplace utensils
|
| 730 |
+
729:sweatshirts
|
| 731 |
+
730:cellular telephone,cellular phone,cellphone,cell,mobile phone
|
| 732 |
+
731:tallboy
|
| 733 |
+
732:stapler
|
| 734 |
+
733:sauna
|
| 735 |
+
734:test tube
|
| 736 |
+
735:palette
|
| 737 |
+
736:shopping carts
|
| 738 |
+
737:tools
|
| 739 |
+
738:push button,push,button
|
| 740 |
+
739:star
|
| 741 |
+
740:roof rack
|
| 742 |
+
741:barbed wire
|
| 743 |
+
742:spray
|
| 744 |
+
743:ear
|
| 745 |
+
744:sponge
|
| 746 |
+
745:racket
|
| 747 |
+
746:tins
|
| 748 |
+
747:eyeglasses
|
| 749 |
+
748:file
|
| 750 |
+
749:scarfs
|
| 751 |
+
750:sugar bowl
|
| 752 |
+
751:flip flop
|
| 753 |
+
752:headstones
|
| 754 |
+
753:laptop bag
|
| 755 |
+
754:leash
|
| 756 |
+
755:climbing frame
|
| 757 |
+
756:suit hanger
|
| 758 |
+
757:floor spotlight
|
| 759 |
+
758:plate rack
|
| 760 |
+
759:sewer
|
| 761 |
+
760:hard drive
|
| 762 |
+
761:sprinkler
|
| 763 |
+
762:tools box
|
| 764 |
+
763:necklace
|
| 765 |
+
764:bulbs
|
| 766 |
+
765:steel industry
|
| 767 |
+
766:club
|
| 768 |
+
767:jack
|
| 769 |
+
768:door bars
|
| 770 |
+
769:control panel,instrument panel,control board,board,panel
|
| 771 |
+
770:hairbrush
|
| 772 |
+
771:napkin holder
|
| 773 |
+
772:office
|
| 774 |
+
773:smoke detector
|
| 775 |
+
774:utensils
|
| 776 |
+
775:apron
|
| 777 |
+
776:scissors
|
| 778 |
+
777:terminal
|
| 779 |
+
778:grinder
|
| 780 |
+
779:entry phone
|
| 781 |
+
780:newspaper stand
|
| 782 |
+
781:pepper shaker
|
| 783 |
+
782:onions
|
| 784 |
+
783:central processing unit,cpu,central processor,processor,mainframe
|
| 785 |
+
784:tape
|
| 786 |
+
785:bat
|
| 787 |
+
786:coaster
|
| 788 |
+
787:calculator
|
| 789 |
+
788:potatoes
|
| 790 |
+
789:luggage rack
|
| 791 |
+
790:salt
|
| 792 |
+
791:street number
|
| 793 |
+
792:viewpoint
|
| 794 |
+
793:sword
|
| 795 |
+
794:cd
|
| 796 |
+
795:rowing machine
|
| 797 |
+
796:plug
|
| 798 |
+
797:andiron,firedog,dog,dog-iron
|
| 799 |
+
798:pepper
|
| 800 |
+
799:tongs
|
| 801 |
+
800:bonfire
|
| 802 |
+
801:dog dish
|
| 803 |
+
802:belt
|
| 804 |
+
803:dumbbells
|
| 805 |
+
804:videocassette recorder,vcr
|
| 806 |
+
805:hook
|
| 807 |
+
806:envelopes
|
| 808 |
+
807:shower faucet
|
| 809 |
+
808:watch
|
| 810 |
+
809:padlock
|
| 811 |
+
810:swimming pool ladder
|
| 812 |
+
811:spanners
|
| 813 |
+
812:gravy boat
|
| 814 |
+
813:notice board
|
| 815 |
+
814:trash bags
|
| 816 |
+
815:fire alarm
|
| 817 |
+
816:ladle
|
| 818 |
+
817:stethoscope
|
| 819 |
+
818:rocket
|
| 820 |
+
819:funnel
|
| 821 |
+
820:bowling pins
|
| 822 |
+
821:valve
|
| 823 |
+
822:thermometer
|
| 824 |
+
823:cups
|
| 825 |
+
824:spice jar
|
| 826 |
+
825:night light
|
| 827 |
+
826:soaps
|
| 828 |
+
827:games table
|
| 829 |
+
828:slotted spoon
|
| 830 |
+
829:reel
|
| 831 |
+
830:scourer
|
| 832 |
+
831:sleeping robe
|
| 833 |
+
832:desk mat
|
| 834 |
+
833:dumbbell
|
| 835 |
+
834:hammer
|
| 836 |
+
835:tie
|
| 837 |
+
836:typewriter
|
| 838 |
+
837:shaker
|
| 839 |
+
838:cheese dish
|
| 840 |
+
839:sea star
|
| 841 |
+
840:racquet
|
| 842 |
+
841:butane gas cylinder
|
| 843 |
+
842:paper weight
|
| 844 |
+
843:shaving brush
|
| 845 |
+
844:sunglasses
|
| 846 |
+
845:gear shift
|
| 847 |
+
846:towel rail
|
| 848 |
+
847:adding machine,totalizer,totaliser
|
frozenseg/data/datasets/cityscapes_with_prompt_eng.txt
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
0:road,railroad
|
| 2 |
+
1:sidewalk,pavement
|
| 3 |
+
2:building,buildings,edifice,edifices,house,ceiling
|
| 4 |
+
3:wall,walls,brick wall,stone wall,tile wall,wood wall
|
| 5 |
+
4:fence,fences
|
| 6 |
+
5:pole,poles
|
| 7 |
+
6:traffic light,traffic lights
|
| 8 |
+
7:traffic sign,stop sign
|
| 9 |
+
8:vegetation,tree,trees,palm tree,bushes
|
| 10 |
+
9:terrain,river,sand,sea,snow,water,mountain,grass,dirt,rock
|
| 11 |
+
10:sky,clouds
|
| 12 |
+
11:person
|
| 13 |
+
12:rider
|
| 14 |
+
13:car,cars
|
| 15 |
+
14:truck,trucks
|
| 16 |
+
15:bus,buses
|
| 17 |
+
16:train,trains,locomotive,locomotives,freight train
|
| 18 |
+
17:motorcycle,motorcycles
|
| 19 |
+
18:bicycle,bicycles,bike,bikes
|
frozenseg/data/datasets/coco_panoptic_with_prompt_eng.txt
ADDED
|
@@ -0,0 +1,201 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
0:invalid_class_id
|
| 2 |
+
1:person,child,girl,boy,woman,man,people,children,girls,boys,women,men,lady,guy,ladies,guys,clothes
|
| 3 |
+
2:bicycle,bicycles,bike,bikes
|
| 4 |
+
3:car,cars
|
| 5 |
+
4:motorcycle,motorcycles
|
| 6 |
+
5:airplane,airplanes
|
| 7 |
+
6:bus,buses
|
| 8 |
+
7:train,trains,locomotive,locomotives,freight train
|
| 9 |
+
8:truck,trucks
|
| 10 |
+
9:boat,boats
|
| 11 |
+
10:traffic light
|
| 12 |
+
11:fire hydrant
|
| 13 |
+
12:invalid_class_id
|
| 14 |
+
13:stop sign
|
| 15 |
+
14:parking meter
|
| 16 |
+
15:bench,benches
|
| 17 |
+
16:bird,birds
|
| 18 |
+
17:cat,cats,kitties,kitty
|
| 19 |
+
18:dog,dogs,puppy,puppies
|
| 20 |
+
19:horse,horses,foal
|
| 21 |
+
20:sheep
|
| 22 |
+
21:cow,cows,calf
|
| 23 |
+
22:elephant,elephants
|
| 24 |
+
23:bear,bears
|
| 25 |
+
24:zebra,zebras
|
| 26 |
+
25:giraffe,giraffes
|
| 27 |
+
26:invalid_class_id
|
| 28 |
+
27:backpack,backpacks
|
| 29 |
+
28:umbrella,umbrellas
|
| 30 |
+
29:invalid_class_id
|
| 31 |
+
30:invalid_class_id
|
| 32 |
+
31:handbag,handbags
|
| 33 |
+
32:tie
|
| 34 |
+
33:suitcase,suitcases
|
| 35 |
+
34:frisbee
|
| 36 |
+
35:skis
|
| 37 |
+
36:snowboard
|
| 38 |
+
37:sports ball
|
| 39 |
+
38:kite,kites
|
| 40 |
+
39:baseball bat
|
| 41 |
+
40:baseball glove
|
| 42 |
+
41:skateboard
|
| 43 |
+
42:surfboard
|
| 44 |
+
43:tennis racket
|
| 45 |
+
44:bottle,bottles,water bottle
|
| 46 |
+
45:invalid_class_id
|
| 47 |
+
46:wine glass,wine glasses,wineglass
|
| 48 |
+
47:cup,cups,water cup,water glass
|
| 49 |
+
48:fork,forks
|
| 50 |
+
49:knife,knives
|
| 51 |
+
50:spoon,spoons
|
| 52 |
+
51:bowl,bowls
|
| 53 |
+
52:banana,bananas
|
| 54 |
+
53:apple,apples,apple fruit
|
| 55 |
+
54:sandwich,sandwiches
|
| 56 |
+
55:orange fruit
|
| 57 |
+
56:broccoli
|
| 58 |
+
57:carrot,carrots
|
| 59 |
+
58:hot dog
|
| 60 |
+
59:pizza
|
| 61 |
+
60:donut,donuts
|
| 62 |
+
61:cake,cakes
|
| 63 |
+
62:chair,chairs
|
| 64 |
+
63:couch,sofa,sofas
|
| 65 |
+
64:potted plant,potted plants,pottedplant,pottedplants,planter,planters
|
| 66 |
+
65:bed,beds
|
| 67 |
+
66:invalid_class_id
|
| 68 |
+
67:dining table,dining tables,diningtable,diningtables,plate,plates,diningtable tablecloth
|
| 69 |
+
68:invalid_class_id
|
| 70 |
+
69:invalid_class_id
|
| 71 |
+
70:toilet
|
| 72 |
+
71:invalid_class_id
|
| 73 |
+
72:tv
|
| 74 |
+
73:laptop
|
| 75 |
+
74:mouse
|
| 76 |
+
75:tv remote,remote control
|
| 77 |
+
76:keyboard
|
| 78 |
+
77:cell phone,mobile
|
| 79 |
+
78:microwave
|
| 80 |
+
79:oven,ovens
|
| 81 |
+
80:toaster
|
| 82 |
+
81:sink,sinks
|
| 83 |
+
82:refrigerator,fridge
|
| 84 |
+
83:invalid_class_id
|
| 85 |
+
84:book,books
|
| 86 |
+
85:clock
|
| 87 |
+
86:vase,vases
|
| 88 |
+
87:scissor,scissors
|
| 89 |
+
88:teddy bear,teddy bears
|
| 90 |
+
89:hair drier
|
| 91 |
+
90:toothbrush,toothbrushes
|
| 92 |
+
91:invalid_class_id
|
| 93 |
+
92:banner,banners
|
| 94 |
+
93:blanket,blankets
|
| 95 |
+
94:invalid_class_id
|
| 96 |
+
95:bridge
|
| 97 |
+
96:invalid_class_id
|
| 98 |
+
97:invalid_class_id
|
| 99 |
+
98:invalid_class_id
|
| 100 |
+
99:invalid_class_id
|
| 101 |
+
100:cardboard
|
| 102 |
+
101:invalid_class_id
|
| 103 |
+
102:invalid_class_id
|
| 104 |
+
103:invalid_class_id
|
| 105 |
+
104:invalid_class_id
|
| 106 |
+
105:invalid_class_id
|
| 107 |
+
106:invalid_class_id
|
| 108 |
+
107:counter
|
| 109 |
+
108:invalid_class_id
|
| 110 |
+
109:curtain,curtains
|
| 111 |
+
110:invalid_class_id
|
| 112 |
+
111:invalid_class_id
|
| 113 |
+
112:door,doors
|
| 114 |
+
113:invalid_class_id
|
| 115 |
+
114:invalid_class_id
|
| 116 |
+
115:invalid_class_id
|
| 117 |
+
116:invalid_class_id
|
| 118 |
+
117:invalid_class_id
|
| 119 |
+
118:wood floor
|
| 120 |
+
119:flower,flowers
|
| 121 |
+
120:invalid_class_id
|
| 122 |
+
121:invalid_class_id
|
| 123 |
+
122:fruit,fruits
|
| 124 |
+
123:invalid_class_id
|
| 125 |
+
124:invalid_class_id
|
| 126 |
+
125:gravel
|
| 127 |
+
126:invalid_class_id
|
| 128 |
+
127:invalid_class_id
|
| 129 |
+
128:house
|
| 130 |
+
129:invalid_class_id
|
| 131 |
+
130:lamp,bulb,lamps,bulbs
|
| 132 |
+
131:invalid_class_id
|
| 133 |
+
132:invalid_class_id
|
| 134 |
+
133:mirror
|
| 135 |
+
134:invalid_class_id
|
| 136 |
+
135:invalid_class_id
|
| 137 |
+
136:invalid_class_id
|
| 138 |
+
137:invalid_class_id
|
| 139 |
+
138:tennis net
|
| 140 |
+
139:invalid_class_id
|
| 141 |
+
140:invalid_class_id
|
| 142 |
+
141:pillow,pillows
|
| 143 |
+
142:invalid_class_id
|
| 144 |
+
143:invalid_class_id
|
| 145 |
+
144:platform
|
| 146 |
+
145:playingfield,tennis court,baseball field,soccer field,tennis field
|
| 147 |
+
146:invalid_class_id
|
| 148 |
+
147:railroad
|
| 149 |
+
148:river
|
| 150 |
+
149:road
|
| 151 |
+
150:invalid_class_id
|
| 152 |
+
151:roof
|
| 153 |
+
152:invalid_class_id
|
| 154 |
+
153:invalid_class_id
|
| 155 |
+
154:sand
|
| 156 |
+
155:sea,sea wave,wave,waves
|
| 157 |
+
156:shelf
|
| 158 |
+
157:invalid_class_id
|
| 159 |
+
158:invalid_class_id
|
| 160 |
+
159:snow
|
| 161 |
+
160:invalid_class_id
|
| 162 |
+
161:stairs
|
| 163 |
+
162:invalid_class_id
|
| 164 |
+
163:invalid_class_id
|
| 165 |
+
164:invalid_class_id
|
| 166 |
+
165:invalid_class_id
|
| 167 |
+
166:tent
|
| 168 |
+
167:invalid_class_id
|
| 169 |
+
168:towel
|
| 170 |
+
169:invalid_class_id
|
| 171 |
+
170:invalid_class_id
|
| 172 |
+
171:brick wall
|
| 173 |
+
172:invalid_class_id
|
| 174 |
+
173:invalid_class_id
|
| 175 |
+
174:invalid_class_id
|
| 176 |
+
175:stone wall
|
| 177 |
+
176:tile wall
|
| 178 |
+
177:wood wall
|
| 179 |
+
178:water
|
| 180 |
+
179:invalid_class_id
|
| 181 |
+
180:window blind
|
| 182 |
+
181:window
|
| 183 |
+
182:invalid_class_id
|
| 184 |
+
183:invalid_class_id
|
| 185 |
+
184:tree,trees,palm tree,bushes
|
| 186 |
+
185:fence,fences
|
| 187 |
+
186:ceiling
|
| 188 |
+
187:sky,clouds
|
| 189 |
+
188:cabinet,cabinets
|
| 190 |
+
189:table
|
| 191 |
+
190:floor,flooring,tile floor
|
| 192 |
+
191:pavement
|
| 193 |
+
192:mountain,mountains
|
| 194 |
+
193:grass
|
| 195 |
+
194:dirt
|
| 196 |
+
195:paper
|
| 197 |
+
196:food
|
| 198 |
+
197:building,buildings
|
| 199 |
+
198:rock
|
| 200 |
+
199:wall,walls
|
| 201 |
+
200:rug
|
frozenseg/data/datasets/coco_stuff_with_prompt_eng.txt
ADDED
|
@@ -0,0 +1,183 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
0:invalid_class_id
|
| 2 |
+
1:person,child,girl,boy,woman,man,people,children,girls,boys,women,men,lady,guy,ladies,guys
|
| 3 |
+
2:bicycle,bicycles,bike,bikes
|
| 4 |
+
3:car,cars
|
| 5 |
+
4:motorcycle,motorcycles
|
| 6 |
+
5:airplane,airplanes
|
| 7 |
+
6:bus,buses
|
| 8 |
+
7:train,trains,locomotive,locomotives,freight train
|
| 9 |
+
8:truck,trucks
|
| 10 |
+
9:boat,boats
|
| 11 |
+
10:traffic light
|
| 12 |
+
11:fire hydrant
|
| 13 |
+
12:invalid_class_id
|
| 14 |
+
13:stop sign
|
| 15 |
+
14:parking meter
|
| 16 |
+
15:bench,benches
|
| 17 |
+
16:bird,birds
|
| 18 |
+
17:cat,cats,kitties,kitty
|
| 19 |
+
18:dog,dogs,puppy,puppies
|
| 20 |
+
19:horse,horses,foal
|
| 21 |
+
20:sheep
|
| 22 |
+
21:cow,cows,calf
|
| 23 |
+
22:elephant,elephants
|
| 24 |
+
23:bear,bears
|
| 25 |
+
24:zebra,zebras
|
| 26 |
+
25:giraffe,giraffes
|
| 27 |
+
26:invalid_class_id
|
| 28 |
+
27:backpack,backpacks
|
| 29 |
+
28:umbrella,umbrellas
|
| 30 |
+
29:invalid_class_id
|
| 31 |
+
30:invalid_class_id
|
| 32 |
+
31:handbag,handbags
|
| 33 |
+
32:tie
|
| 34 |
+
33:suitcase,suitcases
|
| 35 |
+
34:frisbee
|
| 36 |
+
35:skis
|
| 37 |
+
36:snowboard
|
| 38 |
+
37:sports ball
|
| 39 |
+
38:kite,kites
|
| 40 |
+
39:baseball bat
|
| 41 |
+
40:baseball glove
|
| 42 |
+
41:skateboard
|
| 43 |
+
42:surfboard
|
| 44 |
+
43:tennis racket
|
| 45 |
+
44:bottle,bottles,water bottle
|
| 46 |
+
45:invalid_class_id
|
| 47 |
+
46:wine glass,wine glasses,wineglass
|
| 48 |
+
47:cup,cups,water cup,water glass
|
| 49 |
+
48:fork,forks
|
| 50 |
+
49:knife,knives
|
| 51 |
+
50:spoon,spoons
|
| 52 |
+
51:bowl,bowls
|
| 53 |
+
52:banana,bananas
|
| 54 |
+
53:apple,apples,apple fruit
|
| 55 |
+
54:sandwich,sandwiches
|
| 56 |
+
55:orange,oranges,orange fruit
|
| 57 |
+
56:broccoli
|
| 58 |
+
57:carrot,carrots
|
| 59 |
+
58:hot dog
|
| 60 |
+
59:pizza
|
| 61 |
+
60:donut,donuts
|
| 62 |
+
61:cake,cakes
|
| 63 |
+
62:chair,chairs
|
| 64 |
+
63:couch,sofa,sofas
|
| 65 |
+
64:potted plant,potted plants,pottedplant,pottedplants,planter,planters
|
| 66 |
+
65:bed,beds
|
| 67 |
+
66:invalid_class_id
|
| 68 |
+
67:dining table,dining tables,diningtable,diningtables,plate,plates,diningtable tablecloth
|
| 69 |
+
68:invalid_class_id
|
| 70 |
+
69:invalid_class_id
|
| 71 |
+
70:toilet
|
| 72 |
+
71:invalid_class_id
|
| 73 |
+
72:tv
|
| 74 |
+
73:laptop
|
| 75 |
+
74:mouse
|
| 76 |
+
75:remote,tv remote,remote control
|
| 77 |
+
76:keyboard
|
| 78 |
+
77:cell phone,mobile
|
| 79 |
+
78:microwave
|
| 80 |
+
79:oven,ovens
|
| 81 |
+
80:toaster
|
| 82 |
+
81:sink,sinks
|
| 83 |
+
82:refrigerator,fridge
|
| 84 |
+
83:invalid_class_id
|
| 85 |
+
84:book,books
|
| 86 |
+
85:clock
|
| 87 |
+
86:vase,vases
|
| 88 |
+
87:scissors,scissor
|
| 89 |
+
88:teddy bear,teddy bears
|
| 90 |
+
89:hair drier
|
| 91 |
+
90:toothbrush,toothbrushes
|
| 92 |
+
91:invalid_class_id
|
| 93 |
+
92:banner,banners
|
| 94 |
+
93:blanket,blankets
|
| 95 |
+
94:branch
|
| 96 |
+
95:bridge
|
| 97 |
+
96:building,buildings
|
| 98 |
+
97:bush,bushes
|
| 99 |
+
98:cabinet,cabinets
|
| 100 |
+
99:cage,cages
|
| 101 |
+
100:cardboard
|
| 102 |
+
101:carpet,carpets
|
| 103 |
+
102:ceiling-other,ceiling
|
| 104 |
+
103:ceiling-tile,ceiling tile
|
| 105 |
+
104:cloth
|
| 106 |
+
105:clothes
|
| 107 |
+
106:clouds
|
| 108 |
+
107:counter
|
| 109 |
+
108:cupboard,cupboards
|
| 110 |
+
109:curtain,curtains
|
| 111 |
+
110:desk-stuff,desk,desks
|
| 112 |
+
111:dirt
|
| 113 |
+
112:door-stuff,door,doors
|
| 114 |
+
113:fence,fences
|
| 115 |
+
114:floor-marble,marble floor,floor marble
|
| 116 |
+
115:floor-other,floor
|
| 117 |
+
116:floor-stone,stone floor,floor stone
|
| 118 |
+
117:floor-tile,tile floor,floor tile
|
| 119 |
+
118:floor-wood,wood floor,floor wood
|
| 120 |
+
119:flower,flowers
|
| 121 |
+
120:fog
|
| 122 |
+
121:food-other,food
|
| 123 |
+
122:fruit,fruits
|
| 124 |
+
123:furniture-other,furniture
|
| 125 |
+
124:grass
|
| 126 |
+
125:gravel
|
| 127 |
+
126:ground-other,ground
|
| 128 |
+
127:hill
|
| 129 |
+
128:house
|
| 130 |
+
129:leaves
|
| 131 |
+
130:light
|
| 132 |
+
131:mat
|
| 133 |
+
132:metal
|
| 134 |
+
133:mirror-stuff,mirror
|
| 135 |
+
134:moss
|
| 136 |
+
135:mountain,mountains
|
| 137 |
+
136:mud
|
| 138 |
+
137:napkin
|
| 139 |
+
138:net
|
| 140 |
+
139:paper
|
| 141 |
+
140:pavement
|
| 142 |
+
141:pillow,pillows
|
| 143 |
+
142:plant-other
|
| 144 |
+
143:plastic
|
| 145 |
+
144:platform
|
| 146 |
+
145:playingfield,tennis court,baseball field,soccer field,tennis field
|
| 147 |
+
146:railing
|
| 148 |
+
147:railroad
|
| 149 |
+
148:river
|
| 150 |
+
149:road
|
| 151 |
+
150:rock
|
| 152 |
+
151:roof
|
| 153 |
+
152:rug
|
| 154 |
+
153:salad
|
| 155 |
+
154:sand
|
| 156 |
+
155:sea,sea wave,wave,waves
|
| 157 |
+
156:shelf
|
| 158 |
+
157:sky-other,sky
|
| 159 |
+
158:skyscraper
|
| 160 |
+
159:snow
|
| 161 |
+
160:solid-other,solid
|
| 162 |
+
161:stairs
|
| 163 |
+
162:stone
|
| 164 |
+
163:straw
|
| 165 |
+
164:structural-other,structural
|
| 166 |
+
165:table
|
| 167 |
+
166:tent
|
| 168 |
+
167:textile-other,textile
|
| 169 |
+
168:towel
|
| 170 |
+
169:tree,trees,palm tree
|
| 171 |
+
170:vegetable
|
| 172 |
+
171:wall-brick,brick wall,wall brick
|
| 173 |
+
172:wall-concrete,concrete wall,wall concrete
|
| 174 |
+
173:wall-other,wall
|
| 175 |
+
174:wall-panel,wall panel,panel wall
|
| 176 |
+
175:wall-stone,stone wall,wall stone
|
| 177 |
+
176:wall-tile,wall tile,tile wall
|
| 178 |
+
177:wall-wood,wood wall, wall wood
|
| 179 |
+
178:water-other,water
|
| 180 |
+
179:waterdrops
|
| 181 |
+
180:window-blind,window blind
|
| 182 |
+
181:window-other,window
|
| 183 |
+
182:wood
|
frozenseg/data/datasets/lvis_1203_with_prompt_eng.txt
ADDED
|
@@ -0,0 +1,1203 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
1:aerosol can,spray can
|
| 2 |
+
2:air conditioner
|
| 3 |
+
3:airplane,aeroplane
|
| 4 |
+
4:alarm clock
|
| 5 |
+
5:alcohol,alcoholic beverage
|
| 6 |
+
6:alligator,gator
|
| 7 |
+
7:almond
|
| 8 |
+
8:ambulance
|
| 9 |
+
9:amplifier
|
| 10 |
+
10:anklet,ankle bracelet
|
| 11 |
+
11:antenna,aerial,transmitting aerial
|
| 12 |
+
12:apple
|
| 13 |
+
13:applesauce
|
| 14 |
+
14:apricot
|
| 15 |
+
15:apron
|
| 16 |
+
16:aquarium,fish tank
|
| 17 |
+
17:arctic (type of shoe),galosh,golosh,rubber (type of shoe),gumshoe
|
| 18 |
+
18:armband
|
| 19 |
+
19:armchair
|
| 20 |
+
20:armoire
|
| 21 |
+
21:armor,armour
|
| 22 |
+
22:artichoke
|
| 23 |
+
23:trash can,garbage can,wastebin,dustbin,trash barrel,trash bin
|
| 24 |
+
24:ashtray
|
| 25 |
+
25:asparagus
|
| 26 |
+
26:atomizer,atomiser,spray,sprayer,nebulizer,nebuliser
|
| 27 |
+
27:avocado
|
| 28 |
+
28:award,accolade
|
| 29 |
+
29:awning
|
| 30 |
+
30:ax,axe
|
| 31 |
+
31:baboon
|
| 32 |
+
32:baby buggy,baby carriage,perambulator,pram,stroller
|
| 33 |
+
33:basketball backboard
|
| 34 |
+
34:backpack,knapsack,packsack,rucksack,haversack
|
| 35 |
+
35:handbag,purse,pocketbook
|
| 36 |
+
36:suitcase,baggage,luggage
|
| 37 |
+
37:bagel,beigel
|
| 38 |
+
38:bagpipe
|
| 39 |
+
39:baguet,baguette
|
| 40 |
+
40:bait,lure
|
| 41 |
+
41:ball
|
| 42 |
+
42:ballet skirt,tutu
|
| 43 |
+
43:balloon
|
| 44 |
+
44:bamboo
|
| 45 |
+
45:banana
|
| 46 |
+
46:Band Aid
|
| 47 |
+
47:bandage
|
| 48 |
+
48:bandanna,bandana
|
| 49 |
+
49:banjo
|
| 50 |
+
50:banner,streamer
|
| 51 |
+
51:barbell
|
| 52 |
+
52:barge
|
| 53 |
+
53:barrel,cask
|
| 54 |
+
54:barrette
|
| 55 |
+
55:barrow,garden cart,lawn cart,wheelbarrow
|
| 56 |
+
56:baseball base
|
| 57 |
+
57:baseball
|
| 58 |
+
58:baseball bat
|
| 59 |
+
59:baseball cap,jockey cap,golf cap
|
| 60 |
+
60:baseball glove,baseball mitt
|
| 61 |
+
61:basket,handbasket
|
| 62 |
+
62:basketball
|
| 63 |
+
63:bass horn,sousaphone,tuba
|
| 64 |
+
64:bat (animal)
|
| 65 |
+
65:bath mat
|
| 66 |
+
66:bath towel
|
| 67 |
+
67:bathrobe
|
| 68 |
+
68:bathtub,bathing tub
|
| 69 |
+
69:batter (food)
|
| 70 |
+
70:battery
|
| 71 |
+
71:beachball
|
| 72 |
+
72:bead
|
| 73 |
+
73:bean curd,tofu
|
| 74 |
+
74:beanbag
|
| 75 |
+
75:beanie,beany
|
| 76 |
+
76:bear
|
| 77 |
+
77:bed
|
| 78 |
+
78:bedpan
|
| 79 |
+
79:bedspread,bedcover,bed covering,counterpane,spread
|
| 80 |
+
80:cow
|
| 81 |
+
81:beef (food),boeuf (food)
|
| 82 |
+
82:beeper,pager
|
| 83 |
+
83:beer bottle
|
| 84 |
+
84:beer can
|
| 85 |
+
85:beetle
|
| 86 |
+
86:bell
|
| 87 |
+
87:bell pepper,capsicum
|
| 88 |
+
88:belt
|
| 89 |
+
89:belt buckle
|
| 90 |
+
90:bench
|
| 91 |
+
91:beret
|
| 92 |
+
92:bib
|
| 93 |
+
93:Bible
|
| 94 |
+
94:bicycle,bike (bicycle)
|
| 95 |
+
95:visor,vizor
|
| 96 |
+
96:billboard
|
| 97 |
+
97:binder,ring-binder
|
| 98 |
+
98:binoculars,field glasses,opera glasses
|
| 99 |
+
99:bird
|
| 100 |
+
100:birdfeeder
|
| 101 |
+
101:birdbath
|
| 102 |
+
102:birdcage
|
| 103 |
+
103:birdhouse
|
| 104 |
+
104:birthday cake
|
| 105 |
+
105:birthday card
|
| 106 |
+
106:pirate flag
|
| 107 |
+
107:black sheep
|
| 108 |
+
108:blackberry
|
| 109 |
+
109:blackboard,chalkboard
|
| 110 |
+
110:blanket
|
| 111 |
+
111:blazer,sport jacket,sport coat,sports jacket,sports coat
|
| 112 |
+
112:blender,liquidizer,liquidiser
|
| 113 |
+
113:blimp
|
| 114 |
+
114:blinker,flasher
|
| 115 |
+
115:blouse
|
| 116 |
+
116:blueberry
|
| 117 |
+
117:gameboard
|
| 118 |
+
118:boat,ship (boat)
|
| 119 |
+
119:bob,bobber,bobfloat
|
| 120 |
+
120:bobbin,spool,reel
|
| 121 |
+
121:bobby pin,hairgrip
|
| 122 |
+
122:boiled egg,coddled egg
|
| 123 |
+
123:bolo tie,bolo,bola tie,bola
|
| 124 |
+
124:deadbolt
|
| 125 |
+
125:bolt
|
| 126 |
+
126:bonnet
|
| 127 |
+
127:book
|
| 128 |
+
128:bookcase
|
| 129 |
+
129:booklet,brochure,leaflet,pamphlet
|
| 130 |
+
130:bookmark,bookmarker
|
| 131 |
+
131:boom microphone,microphone boom
|
| 132 |
+
132:boot
|
| 133 |
+
133:bottle
|
| 134 |
+
134:bottle opener
|
| 135 |
+
135:bouquet
|
| 136 |
+
136:bow (weapon)
|
| 137 |
+
137:bow (decorative ribbons)
|
| 138 |
+
138:bow-tie,bowtie
|
| 139 |
+
139:bowl
|
| 140 |
+
140:pipe bowl
|
| 141 |
+
141:bowler hat,bowler,derby hat,derby,plug hat
|
| 142 |
+
142:bowling ball
|
| 143 |
+
143:box
|
| 144 |
+
144:boxing glove
|
| 145 |
+
145:suspenders
|
| 146 |
+
146:bracelet,bangle
|
| 147 |
+
147:brass plaque
|
| 148 |
+
148:brassiere,bra,bandeau
|
| 149 |
+
149:bread-bin,breadbox
|
| 150 |
+
150:bread
|
| 151 |
+
151:breechcloth,breechclout,loincloth
|
| 152 |
+
152:bridal gown,wedding gown,wedding dress
|
| 153 |
+
153:briefcase
|
| 154 |
+
154:broccoli
|
| 155 |
+
155:broach
|
| 156 |
+
156:broom
|
| 157 |
+
157:brownie
|
| 158 |
+
158:brussels sprouts
|
| 159 |
+
159:bubble gum
|
| 160 |
+
160:bucket,pail
|
| 161 |
+
161:horse buggy
|
| 162 |
+
162:horned cow
|
| 163 |
+
163:bulldog
|
| 164 |
+
164:bulldozer,dozer
|
| 165 |
+
165:bullet train
|
| 166 |
+
166:bulletin board,notice board
|
| 167 |
+
167:bulletproof vest
|
| 168 |
+
168:bullhorn,megaphone
|
| 169 |
+
169:bun,roll
|
| 170 |
+
170:bunk bed
|
| 171 |
+
171:buoy
|
| 172 |
+
172:burrito
|
| 173 |
+
173:bus (vehicle),autobus,charabanc,double-decker,motorbus,motorcoach
|
| 174 |
+
174:business card
|
| 175 |
+
175:butter
|
| 176 |
+
176:butterfly
|
| 177 |
+
177:button
|
| 178 |
+
178:cab (taxi),taxi,taxicab
|
| 179 |
+
179:cabana
|
| 180 |
+
180:cabin car,caboose
|
| 181 |
+
181:cabinet
|
| 182 |
+
182:locker,storage locker
|
| 183 |
+
183:cake
|
| 184 |
+
184:calculator
|
| 185 |
+
185:calendar
|
| 186 |
+
186:calf
|
| 187 |
+
187:camcorder
|
| 188 |
+
188:camel
|
| 189 |
+
189:camera
|
| 190 |
+
190:camera lens
|
| 191 |
+
191:camper (vehicle),camping bus,motor home
|
| 192 |
+
192:can,tin can
|
| 193 |
+
193:can opener,tin opener
|
| 194 |
+
194:candle,candlestick
|
| 195 |
+
195:candle holder
|
| 196 |
+
196:candy bar
|
| 197 |
+
197:candy cane
|
| 198 |
+
198:walking cane
|
| 199 |
+
199:canister,cannister
|
| 200 |
+
200:canoe
|
| 201 |
+
201:cantaloup,cantaloupe
|
| 202 |
+
202:canteen
|
| 203 |
+
203:cap (headwear)
|
| 204 |
+
204:bottle cap,cap (container lid)
|
| 205 |
+
205:cape
|
| 206 |
+
206:cappuccino,coffee cappuccino
|
| 207 |
+
207:car (automobile),auto (automobile),automobile
|
| 208 |
+
208:railcar (part of a train),railway car (part of a train),railroad car (part of a train)
|
| 209 |
+
209:elevator car
|
| 210 |
+
210:car battery,automobile battery
|
| 211 |
+
211:identity card
|
| 212 |
+
212:card
|
| 213 |
+
213:cardigan
|
| 214 |
+
214:cargo ship,cargo vessel
|
| 215 |
+
215:carnation
|
| 216 |
+
216:horse carriage
|
| 217 |
+
217:carrot
|
| 218 |
+
218:tote bag
|
| 219 |
+
219:cart
|
| 220 |
+
220:carton
|
| 221 |
+
221:cash register,register (for cash transactions)
|
| 222 |
+
222:casserole
|
| 223 |
+
223:cassette
|
| 224 |
+
224:cast,plaster cast,plaster bandage
|
| 225 |
+
225:cat
|
| 226 |
+
226:cauliflower
|
| 227 |
+
227:cayenne (spice),cayenne pepper (spice),red pepper (spice)
|
| 228 |
+
228:CD player
|
| 229 |
+
229:celery
|
| 230 |
+
230:cellular telephone,cellular phone,cellphone,mobile phone,smart phone
|
| 231 |
+
231:chain mail,ring mail,chain armor,chain armour,ring armor,ring armour
|
| 232 |
+
232:chair
|
| 233 |
+
233:chaise longue,chaise,daybed
|
| 234 |
+
234:chalice
|
| 235 |
+
235:chandelier
|
| 236 |
+
236:chap
|
| 237 |
+
237:checkbook,chequebook
|
| 238 |
+
238:checkerboard
|
| 239 |
+
239:cherry
|
| 240 |
+
240:chessboard
|
| 241 |
+
241:chicken (animal)
|
| 242 |
+
242:chickpea,garbanzo
|
| 243 |
+
243:chili (vegetable),chili pepper (vegetable),chilli (vegetable),chilly (vegetable),chile (vegetable)
|
| 244 |
+
244:chime,gong
|
| 245 |
+
245:chinaware
|
| 246 |
+
246:crisp (potato chip),potato chip
|
| 247 |
+
247:poker chip
|
| 248 |
+
248:chocolate bar
|
| 249 |
+
249:chocolate cake
|
| 250 |
+
250:chocolate milk
|
| 251 |
+
251:chocolate mousse
|
| 252 |
+
252:choker,collar,neckband
|
| 253 |
+
253:chopping board,cutting board,chopping block
|
| 254 |
+
254:chopstick
|
| 255 |
+
255:Christmas tree
|
| 256 |
+
256:slide
|
| 257 |
+
257:cider,cyder
|
| 258 |
+
258:cigar box
|
| 259 |
+
259:cigarette
|
| 260 |
+
260:cigarette case,cigarette pack
|
| 261 |
+
261:cistern,water tank
|
| 262 |
+
262:clarinet
|
| 263 |
+
263:clasp
|
| 264 |
+
264:cleansing agent,cleanser,cleaner
|
| 265 |
+
265:cleat (for securing rope)
|
| 266 |
+
266:clementine
|
| 267 |
+
267:clip
|
| 268 |
+
268:clipboard
|
| 269 |
+
269:clippers (for plants)
|
| 270 |
+
270:cloak
|
| 271 |
+
271:clock,timepiece,timekeeper
|
| 272 |
+
272:clock tower
|
| 273 |
+
273:clothes hamper,laundry basket,clothes basket
|
| 274 |
+
274:clothespin,clothes peg
|
| 275 |
+
275:clutch bag
|
| 276 |
+
276:coaster
|
| 277 |
+
277:coat
|
| 278 |
+
278:coat hanger,clothes hanger,dress hanger
|
| 279 |
+
279:coatrack,hatrack
|
| 280 |
+
280:cock,rooster
|
| 281 |
+
281:cockroach
|
| 282 |
+
282:cocoa (beverage),hot chocolate (beverage),drinking chocolate
|
| 283 |
+
283:coconut,cocoanut
|
| 284 |
+
284:coffee maker,coffee machine
|
| 285 |
+
285:coffee table,cocktail table
|
| 286 |
+
286:coffeepot
|
| 287 |
+
287:coil
|
| 288 |
+
288:coin
|
| 289 |
+
289:colander,cullender
|
| 290 |
+
290:coleslaw,slaw
|
| 291 |
+
291:coloring material,colouring material
|
| 292 |
+
292:combination lock
|
| 293 |
+
293:pacifier,teething ring
|
| 294 |
+
294:comic book
|
| 295 |
+
295:compass
|
| 296 |
+
296:computer keyboard,keyboard (computer)
|
| 297 |
+
297:condiment
|
| 298 |
+
298:cone,traffic cone
|
| 299 |
+
299:control,controller
|
| 300 |
+
300:convertible (automobile)
|
| 301 |
+
301:sofa bed
|
| 302 |
+
302:cooker
|
| 303 |
+
303:cookie,cooky,biscuit (cookie)
|
| 304 |
+
304:cooking utensil
|
| 305 |
+
305:cooler (for food),ice chest
|
| 306 |
+
306:cork (bottle plug),bottle cork
|
| 307 |
+
307:corkboard
|
| 308 |
+
308:corkscrew,bottle screw
|
| 309 |
+
309:edible corn,corn,maize
|
| 310 |
+
310:cornbread
|
| 311 |
+
311:cornet,horn,trumpet
|
| 312 |
+
312:cornice,valance,valance board,pelmet
|
| 313 |
+
313:cornmeal
|
| 314 |
+
314:corset,girdle
|
| 315 |
+
315:costume
|
| 316 |
+
316:cougar,puma,catamount,mountain lion,panther
|
| 317 |
+
317:coverall
|
| 318 |
+
318:cowbell
|
| 319 |
+
319:cowboy hat,ten-gallon hat
|
| 320 |
+
320:crab (animal)
|
| 321 |
+
321:crabmeat
|
| 322 |
+
322:cracker
|
| 323 |
+
323:crape,crepe,French pancake
|
| 324 |
+
324:crate
|
| 325 |
+
325:crayon,wax crayon
|
| 326 |
+
326:cream pitcher
|
| 327 |
+
327:crescent roll,croissant
|
| 328 |
+
328:crib,cot
|
| 329 |
+
329:crock pot,earthenware jar
|
| 330 |
+
330:crossbar
|
| 331 |
+
331:crouton
|
| 332 |
+
332:crow
|
| 333 |
+
333:crowbar,wrecking bar,pry bar
|
| 334 |
+
334:crown
|
| 335 |
+
335:crucifix
|
| 336 |
+
336:cruise ship,cruise liner
|
| 337 |
+
337:police cruiser,patrol car,police car,squad car
|
| 338 |
+
338:crumb
|
| 339 |
+
339:crutch
|
| 340 |
+
340:cub (animal)
|
| 341 |
+
341:cube,square block
|
| 342 |
+
342:cucumber,cuke
|
| 343 |
+
343:cufflink
|
| 344 |
+
344:cup
|
| 345 |
+
345:trophy cup
|
| 346 |
+
346:cupboard,closet
|
| 347 |
+
347:cupcake
|
| 348 |
+
348:hair curler,hair roller,hair crimper
|
| 349 |
+
349:curling iron
|
| 350 |
+
350:curtain,drapery
|
| 351 |
+
351:cushion
|
| 352 |
+
352:cylinder
|
| 353 |
+
353:cymbal
|
| 354 |
+
354:dagger
|
| 355 |
+
355:dalmatian
|
| 356 |
+
356:dartboard
|
| 357 |
+
357:date (fruit)
|
| 358 |
+
358:deck chair,beach chair
|
| 359 |
+
359:deer,cervid
|
| 360 |
+
360:dental floss,floss
|
| 361 |
+
361:desk
|
| 362 |
+
362:detergent
|
| 363 |
+
363:diaper
|
| 364 |
+
364:diary,journal
|
| 365 |
+
365:die,dice
|
| 366 |
+
366:dinghy,dory,rowboat
|
| 367 |
+
367:dining table
|
| 368 |
+
368:tux,tuxedo
|
| 369 |
+
369:dish
|
| 370 |
+
370:dish antenna
|
| 371 |
+
371:dishrag,dishcloth
|
| 372 |
+
372:dishtowel,tea towel
|
| 373 |
+
373:dishwasher,dishwashing machine
|
| 374 |
+
374:dishwasher detergent,dishwashing detergent,dishwashing liquid,dishsoap
|
| 375 |
+
375:dispenser
|
| 376 |
+
376:diving board
|
| 377 |
+
377:Dixie cup,paper cup
|
| 378 |
+
378:dog
|
| 379 |
+
379:dog collar
|
| 380 |
+
380:doll
|
| 381 |
+
381:dollar,dollar bill,one dollar bill
|
| 382 |
+
382:dollhouse,doll's house
|
| 383 |
+
383:dolphin
|
| 384 |
+
384:domestic ass,donkey
|
| 385 |
+
385:doorknob,doorhandle
|
| 386 |
+
386:doormat,welcome mat
|
| 387 |
+
387:doughnut,donut
|
| 388 |
+
388:dove
|
| 389 |
+
389:dragonfly
|
| 390 |
+
390:drawer
|
| 391 |
+
391:underdrawers,boxers,boxershorts
|
| 392 |
+
392:dress,frock
|
| 393 |
+
393:dress hat,high hat,opera hat,silk hat,top hat
|
| 394 |
+
394:dress suit
|
| 395 |
+
395:dresser
|
| 396 |
+
396:drill
|
| 397 |
+
397:drone
|
| 398 |
+
398:dropper,eye dropper
|
| 399 |
+
399:drum (musical instrument)
|
| 400 |
+
400:drumstick
|
| 401 |
+
401:duck
|
| 402 |
+
402:duckling
|
| 403 |
+
403:duct tape
|
| 404 |
+
404:duffel bag,duffle bag,duffel,duffle
|
| 405 |
+
405:dumbbell
|
| 406 |
+
406:dumpster
|
| 407 |
+
407:dustpan
|
| 408 |
+
408:eagle
|
| 409 |
+
409:earphone,earpiece,headphone
|
| 410 |
+
410:earplug
|
| 411 |
+
411:earring
|
| 412 |
+
412:easel
|
| 413 |
+
413:eclair
|
| 414 |
+
414:eel
|
| 415 |
+
415:egg,eggs
|
| 416 |
+
416:egg roll,spring roll
|
| 417 |
+
417:egg yolk,yolk (egg)
|
| 418 |
+
418:eggbeater,eggwhisk
|
| 419 |
+
419:eggplant,aubergine
|
| 420 |
+
420:electric chair
|
| 421 |
+
421:refrigerator
|
| 422 |
+
422:elephant
|
| 423 |
+
423:elk,moose
|
| 424 |
+
424:envelope
|
| 425 |
+
425:eraser
|
| 426 |
+
426:escargot
|
| 427 |
+
427:eyepatch
|
| 428 |
+
428:falcon
|
| 429 |
+
429:fan
|
| 430 |
+
430:faucet,spigot,tap
|
| 431 |
+
431:fedora
|
| 432 |
+
432:ferret
|
| 433 |
+
433:Ferris wheel
|
| 434 |
+
434:ferry,ferryboat
|
| 435 |
+
435:fig (fruit)
|
| 436 |
+
436:fighter jet,fighter aircraft,attack aircraft
|
| 437 |
+
437:figurine
|
| 438 |
+
438:file cabinet,filing cabinet
|
| 439 |
+
439:file (tool)
|
| 440 |
+
440:fire alarm,smoke alarm
|
| 441 |
+
441:fire engine,fire truck
|
| 442 |
+
442:fire extinguisher,extinguisher
|
| 443 |
+
443:fire hose
|
| 444 |
+
444:fireplace
|
| 445 |
+
445:fireplug,fire hydrant,hydrant
|
| 446 |
+
446:first-aid kit
|
| 447 |
+
447:fish
|
| 448 |
+
448:fish (food)
|
| 449 |
+
449:fishbowl,goldfish bowl
|
| 450 |
+
450:fishing rod,fishing pole
|
| 451 |
+
451:flag
|
| 452 |
+
452:flagpole,flagstaff
|
| 453 |
+
453:flamingo
|
| 454 |
+
454:flannel
|
| 455 |
+
455:flap
|
| 456 |
+
456:flash,flashbulb
|
| 457 |
+
457:flashlight,torch
|
| 458 |
+
458:fleece
|
| 459 |
+
459:flip-flop (sandal)
|
| 460 |
+
460:flipper (footwear),fin (footwear)
|
| 461 |
+
461:flower arrangement,floral arrangement
|
| 462 |
+
462:flute glass,champagne flute
|
| 463 |
+
463:foal
|
| 464 |
+
464:folding chair
|
| 465 |
+
465:food processor
|
| 466 |
+
466:football (American)
|
| 467 |
+
467:football helmet
|
| 468 |
+
468:footstool,footrest
|
| 469 |
+
469:fork
|
| 470 |
+
470:forklift
|
| 471 |
+
471:freight car
|
| 472 |
+
472:French toast
|
| 473 |
+
473:freshener,air freshener
|
| 474 |
+
474:frisbee
|
| 475 |
+
475:frog,toad,toad frog
|
| 476 |
+
476:fruit juice
|
| 477 |
+
477:frying pan,frypan,skillet
|
| 478 |
+
478:fudge
|
| 479 |
+
479:funnel
|
| 480 |
+
480:futon
|
| 481 |
+
481:gag,muzzle
|
| 482 |
+
482:garbage
|
| 483 |
+
483:garbage truck
|
| 484 |
+
484:garden hose
|
| 485 |
+
485:gargle,mouthwash
|
| 486 |
+
486:gargoyle
|
| 487 |
+
487:garlic,ail
|
| 488 |
+
488:gasmask,respirator,gas helmet
|
| 489 |
+
489:gazelle
|
| 490 |
+
490:gelatin,jelly
|
| 491 |
+
491:gemstone
|
| 492 |
+
492:generator
|
| 493 |
+
493:giant panda,panda,panda bear
|
| 494 |
+
494:gift wrap
|
| 495 |
+
495:ginger,gingerroot
|
| 496 |
+
496:giraffe
|
| 497 |
+
497:cincture,sash,waistband,waistcloth
|
| 498 |
+
498:glass (drink container),drinking glass
|
| 499 |
+
499:globe
|
| 500 |
+
500:glove
|
| 501 |
+
501:goat
|
| 502 |
+
502:goggles
|
| 503 |
+
503:goldfish
|
| 504 |
+
504:golf club,golf-club
|
| 505 |
+
505:golfcart
|
| 506 |
+
506:gondola (boat)
|
| 507 |
+
507:goose
|
| 508 |
+
508:gorilla
|
| 509 |
+
509:gourd
|
| 510 |
+
510:grape
|
| 511 |
+
511:grater
|
| 512 |
+
512:gravestone,headstone,tombstone
|
| 513 |
+
513:gravy boat,gravy holder
|
| 514 |
+
514:green bean
|
| 515 |
+
515:green onion,spring onion,scallion
|
| 516 |
+
516:griddle
|
| 517 |
+
517:grill,grille,grillwork,radiator grille
|
| 518 |
+
518:grits,hominy grits
|
| 519 |
+
519:grizzly,grizzly bear
|
| 520 |
+
520:grocery bag
|
| 521 |
+
521:guitar
|
| 522 |
+
522:gull,seagull
|
| 523 |
+
523:gun
|
| 524 |
+
524:hairbrush
|
| 525 |
+
525:hairnet
|
| 526 |
+
526:hairpin
|
| 527 |
+
527:halter top
|
| 528 |
+
528:ham,jambon,gammon
|
| 529 |
+
529:hamburger,beefburger,burger
|
| 530 |
+
530:hammer
|
| 531 |
+
531:hammock
|
| 532 |
+
532:hamper
|
| 533 |
+
533:hamster
|
| 534 |
+
534:hair dryer
|
| 535 |
+
535:hand glass,hand mirror
|
| 536 |
+
536:hand towel,face towel
|
| 537 |
+
537:handcart,pushcart,hand truck
|
| 538 |
+
538:handcuff
|
| 539 |
+
539:handkerchief
|
| 540 |
+
540:handle,grip,handgrip
|
| 541 |
+
541:handsaw,carpenter's saw
|
| 542 |
+
542:hardback book,hardcover book
|
| 543 |
+
543:harmonium,organ (musical instrument),reed organ (musical instrument)
|
| 544 |
+
544:hat
|
| 545 |
+
545:hatbox
|
| 546 |
+
546:veil
|
| 547 |
+
547:headband
|
| 548 |
+
548:headboard
|
| 549 |
+
549:headlight,headlamp
|
| 550 |
+
550:headscarf
|
| 551 |
+
551:headset
|
| 552 |
+
552:headstall (for horses),headpiece (for horses)
|
| 553 |
+
553:heart
|
| 554 |
+
554:heater,warmer
|
| 555 |
+
555:helicopter
|
| 556 |
+
556:helmet
|
| 557 |
+
557:heron
|
| 558 |
+
558:highchair,feeding chair
|
| 559 |
+
559:hinge
|
| 560 |
+
560:hippopotamus
|
| 561 |
+
561:hockey stick
|
| 562 |
+
562:hog,pig
|
| 563 |
+
563:home plate (baseball),home base (baseball)
|
| 564 |
+
564:honey
|
| 565 |
+
565:fume hood,exhaust hood
|
| 566 |
+
566:hook
|
| 567 |
+
567:hookah,narghile,nargileh,sheesha,shisha,water pipe
|
| 568 |
+
568:hornet
|
| 569 |
+
569:horse
|
| 570 |
+
570:hose,hosepipe
|
| 571 |
+
571:hot-air balloon
|
| 572 |
+
572:hotplate
|
| 573 |
+
573:hot sauce
|
| 574 |
+
574:hourglass
|
| 575 |
+
575:houseboat
|
| 576 |
+
576:hummingbird
|
| 577 |
+
577:hummus,humus,hommos,hoummos,humous
|
| 578 |
+
578:polar bear
|
| 579 |
+
579:icecream
|
| 580 |
+
580:popsicle
|
| 581 |
+
581:ice maker
|
| 582 |
+
582:ice pack,ice bag
|
| 583 |
+
583:ice skate
|
| 584 |
+
584:igniter,ignitor,lighter
|
| 585 |
+
585:inhaler,inhalator
|
| 586 |
+
586:iPod
|
| 587 |
+
587:iron (for clothing),smoothing iron (for clothing)
|
| 588 |
+
588:ironing board
|
| 589 |
+
589:jacket
|
| 590 |
+
590:jam
|
| 591 |
+
591:jar
|
| 592 |
+
592:jean,blue jean,denim
|
| 593 |
+
593:jeep,landrover
|
| 594 |
+
594:jelly bean,jelly egg
|
| 595 |
+
595:jersey,T-shirt,tee shirt
|
| 596 |
+
596:jet plane,jet-propelled plane
|
| 597 |
+
597:jewel,gem,precious stone
|
| 598 |
+
598:jewelry,jewellery
|
| 599 |
+
599:joystick
|
| 600 |
+
600:jumpsuit
|
| 601 |
+
601:kayak
|
| 602 |
+
602:keg
|
| 603 |
+
603:kennel,doghouse
|
| 604 |
+
604:kettle,boiler
|
| 605 |
+
605:key
|
| 606 |
+
606:keycard
|
| 607 |
+
607:kilt
|
| 608 |
+
608:kimono
|
| 609 |
+
609:kitchen sink
|
| 610 |
+
610:kitchen table
|
| 611 |
+
611:kite
|
| 612 |
+
612:kitten,kitty
|
| 613 |
+
613:kiwi fruit
|
| 614 |
+
614:knee pad
|
| 615 |
+
615:knife
|
| 616 |
+
616:knitting needle
|
| 617 |
+
617:knob
|
| 618 |
+
618:knocker (on a door),doorknocker
|
| 619 |
+
619:koala,koala bear
|
| 620 |
+
620:lab coat,laboratory coat
|
| 621 |
+
621:ladder
|
| 622 |
+
622:ladle
|
| 623 |
+
623:ladybug,ladybeetle,ladybird beetle
|
| 624 |
+
624:lamb (animal)
|
| 625 |
+
625:lamb-chop,lambchop
|
| 626 |
+
626:lamp
|
| 627 |
+
627:lamppost
|
| 628 |
+
628:lampshade
|
| 629 |
+
629:lantern
|
| 630 |
+
630:lanyard,laniard
|
| 631 |
+
631:laptop computer,notebook computer
|
| 632 |
+
632:lasagna,lasagne
|
| 633 |
+
633:latch
|
| 634 |
+
634:lawn mower
|
| 635 |
+
635:leather
|
| 636 |
+
636:legging (clothing),leging (clothing),leg covering
|
| 637 |
+
637:Lego,Lego set
|
| 638 |
+
638:legume
|
| 639 |
+
639:lemon
|
| 640 |
+
640:lemonade
|
| 641 |
+
641:lettuce
|
| 642 |
+
642:license plate,numberplate
|
| 643 |
+
643:life buoy,lifesaver,life belt,life ring
|
| 644 |
+
644:life jacket,life vest
|
| 645 |
+
645:lightbulb
|
| 646 |
+
646:lightning rod,lightning conductor
|
| 647 |
+
647:lime
|
| 648 |
+
648:limousine
|
| 649 |
+
649:lion
|
| 650 |
+
650:lip balm
|
| 651 |
+
651:liquor,spirits,hard liquor,liqueur,cordial
|
| 652 |
+
652:lizard
|
| 653 |
+
653:log
|
| 654 |
+
654:lollipop
|
| 655 |
+
655:speaker (stero equipment)
|
| 656 |
+
656:loveseat
|
| 657 |
+
657:machine gun
|
| 658 |
+
658:magazine
|
| 659 |
+
659:magnet
|
| 660 |
+
660:mail slot
|
| 661 |
+
661:mailbox (at home),letter box (at home)
|
| 662 |
+
662:mallard
|
| 663 |
+
663:mallet
|
| 664 |
+
664:mammoth
|
| 665 |
+
665:manatee
|
| 666 |
+
666:mandarin orange
|
| 667 |
+
667:manger,trough
|
| 668 |
+
668:manhole
|
| 669 |
+
669:map
|
| 670 |
+
670:marker
|
| 671 |
+
671:martini
|
| 672 |
+
672:mascot
|
| 673 |
+
673:mashed potato
|
| 674 |
+
674:masher
|
| 675 |
+
675:mask,facemask
|
| 676 |
+
676:mast
|
| 677 |
+
677:mat (gym equipment),gym mat
|
| 678 |
+
678:matchbox
|
| 679 |
+
679:mattress
|
| 680 |
+
680:measuring cup
|
| 681 |
+
681:measuring stick,ruler (measuring stick),measuring rod
|
| 682 |
+
682:meatball
|
| 683 |
+
683:medicine
|
| 684 |
+
684:melon
|
| 685 |
+
685:microphone
|
| 686 |
+
686:microscope
|
| 687 |
+
687:microwave oven
|
| 688 |
+
688:milestone,milepost
|
| 689 |
+
689:milk
|
| 690 |
+
690:milk can
|
| 691 |
+
691:milkshake
|
| 692 |
+
692:minivan
|
| 693 |
+
693:mint candy
|
| 694 |
+
694:mirror
|
| 695 |
+
695:mitten
|
| 696 |
+
696:mixer (kitchen tool),stand mixer
|
| 697 |
+
697:money
|
| 698 |
+
698:monitor (computer equipment) computer monitor
|
| 699 |
+
699:monkey
|
| 700 |
+
700:motor
|
| 701 |
+
701:motor scooter,scooter
|
| 702 |
+
702:motor vehicle,automotive vehicle
|
| 703 |
+
703:motorcycle
|
| 704 |
+
704:mound (baseball),pitcher's mound
|
| 705 |
+
705:mouse (computer equipment),computer mouse
|
| 706 |
+
706:mousepad
|
| 707 |
+
707:muffin
|
| 708 |
+
708:mug
|
| 709 |
+
709:mushroom
|
| 710 |
+
710:music stool,piano stool
|
| 711 |
+
711:musical instrument,instrument (musical)
|
| 712 |
+
712:nailfile
|
| 713 |
+
713:napkin,table napkin,serviette
|
| 714 |
+
714:neckerchief
|
| 715 |
+
715:necklace
|
| 716 |
+
716:necktie,tie (necktie)
|
| 717 |
+
717:needle
|
| 718 |
+
718:nest
|
| 719 |
+
719:newspaper,paper (newspaper)
|
| 720 |
+
720:newsstand
|
| 721 |
+
721:nightshirt,nightwear,sleepwear,nightclothes
|
| 722 |
+
722:nosebag (for animals),feedbag
|
| 723 |
+
723:noseband (for animals),nosepiece (for animals)
|
| 724 |
+
724:notebook
|
| 725 |
+
725:notepad
|
| 726 |
+
726:nut
|
| 727 |
+
727:nutcracker
|
| 728 |
+
728:oar
|
| 729 |
+
729:octopus (food)
|
| 730 |
+
730:octopus (animal)
|
| 731 |
+
731:oil lamp,kerosene lamp,kerosine lamp
|
| 732 |
+
732:olive oil
|
| 733 |
+
733:omelet,omelette
|
| 734 |
+
734:onion
|
| 735 |
+
735:orange (fruit)
|
| 736 |
+
736:orange juice
|
| 737 |
+
737:ostrich
|
| 738 |
+
738:ottoman,pouf,pouffe,hassock
|
| 739 |
+
739:oven
|
| 740 |
+
740:overalls (clothing)
|
| 741 |
+
741:owl
|
| 742 |
+
742:packet
|
| 743 |
+
743:inkpad,inking pad,stamp pad
|
| 744 |
+
744:pad
|
| 745 |
+
745:paddle,boat paddle
|
| 746 |
+
746:padlock
|
| 747 |
+
747:paintbrush
|
| 748 |
+
748:painting
|
| 749 |
+
749:pajamas,pyjamas
|
| 750 |
+
750:palette,pallet
|
| 751 |
+
751:pan (for cooking),cooking pan
|
| 752 |
+
752:pan (metal container)
|
| 753 |
+
753:pancake
|
| 754 |
+
754:pantyhose
|
| 755 |
+
755:papaya
|
| 756 |
+
756:paper plate
|
| 757 |
+
757:paper towel
|
| 758 |
+
758:paperback book,paper-back book,softback book,soft-cover book
|
| 759 |
+
759:paperweight
|
| 760 |
+
760:parachute
|
| 761 |
+
761:parakeet,parrakeet,parroket,paraquet,paroquet,parroquet
|
| 762 |
+
762:parasail (sports)
|
| 763 |
+
763:parasol,sunshade
|
| 764 |
+
764:parchment
|
| 765 |
+
765:parka,anorak
|
| 766 |
+
766:parking meter
|
| 767 |
+
767:parrot
|
| 768 |
+
768:passenger car (part of a train),coach (part of a train)
|
| 769 |
+
769:passenger ship
|
| 770 |
+
770:passport
|
| 771 |
+
771:pastry
|
| 772 |
+
772:patty (food)
|
| 773 |
+
773:pea (food)
|
| 774 |
+
774:peach
|
| 775 |
+
775:peanut butter
|
| 776 |
+
776:pear
|
| 777 |
+
777:peeler (tool for fruit and vegetables)
|
| 778 |
+
778:wooden leg,pegleg
|
| 779 |
+
779:pegboard
|
| 780 |
+
780:pelican
|
| 781 |
+
781:pen
|
| 782 |
+
782:pencil
|
| 783 |
+
783:pencil box,pencil case
|
| 784 |
+
784:pencil sharpener
|
| 785 |
+
785:pendulum
|
| 786 |
+
786:penguin
|
| 787 |
+
787:pennant
|
| 788 |
+
788:penny (coin)
|
| 789 |
+
789:pepper,peppercorn
|
| 790 |
+
790:pepper mill,pepper grinder
|
| 791 |
+
791:perfume
|
| 792 |
+
792:persimmon
|
| 793 |
+
793:person,baby,child,boy,girl,man,woman,human
|
| 794 |
+
794:pet
|
| 795 |
+
795:pew (church bench),church bench
|
| 796 |
+
796:phonebook,telephone book,telephone directory
|
| 797 |
+
797:phonograph record,phonograph recording,record (phonograph recording)
|
| 798 |
+
798:piano
|
| 799 |
+
799:pickle
|
| 800 |
+
800:pickup truck
|
| 801 |
+
801:pie
|
| 802 |
+
802:pigeon
|
| 803 |
+
803:piggy bank,penny bank
|
| 804 |
+
804:pillow
|
| 805 |
+
805:pin (non jewelry)
|
| 806 |
+
806:pineapple
|
| 807 |
+
807:pinecone
|
| 808 |
+
808:ping-pong ball
|
| 809 |
+
809:pinwheel
|
| 810 |
+
810:tobacco pipe
|
| 811 |
+
811:pipe,piping
|
| 812 |
+
812:pistol,handgun
|
| 813 |
+
813:pita (bread),pocket bread
|
| 814 |
+
814:pitcher (vessel for liquid),ewer
|
| 815 |
+
815:pitchfork
|
| 816 |
+
816:pizza
|
| 817 |
+
817:place mat
|
| 818 |
+
818:plate
|
| 819 |
+
819:platter
|
| 820 |
+
820:playpen
|
| 821 |
+
821:pliers,plyers
|
| 822 |
+
822:plow (farm equipment),plough (farm equipment)
|
| 823 |
+
823:plume
|
| 824 |
+
824:pocket watch
|
| 825 |
+
825:pocketknife
|
| 826 |
+
826:poker (fire stirring tool),stove poker,fire hook
|
| 827 |
+
827:pole,post
|
| 828 |
+
828:polo shirt,sport shirt
|
| 829 |
+
829:poncho
|
| 830 |
+
830:pony
|
| 831 |
+
831:pool table,billiard table,snooker table
|
| 832 |
+
832:pop (soda),soda (pop),tonic,soft drink
|
| 833 |
+
833:postbox (public),mailbox (public)
|
| 834 |
+
834:postcard,postal card,mailing-card
|
| 835 |
+
835:poster,placard
|
| 836 |
+
836:pot
|
| 837 |
+
837:flowerpot
|
| 838 |
+
838:potato
|
| 839 |
+
839:potholder
|
| 840 |
+
840:pottery,clayware
|
| 841 |
+
841:pouch
|
| 842 |
+
842:power shovel,excavator,digger
|
| 843 |
+
843:prawn,shrimp
|
| 844 |
+
844:pretzel
|
| 845 |
+
845:printer,printing machine
|
| 846 |
+
846:projectile (weapon),missile
|
| 847 |
+
847:projector
|
| 848 |
+
848:propeller,propellor
|
| 849 |
+
849:prune
|
| 850 |
+
850:pudding
|
| 851 |
+
851:puffer (fish),pufferfish,blowfish,globefish
|
| 852 |
+
852:puffin
|
| 853 |
+
853:pug-dog
|
| 854 |
+
854:pumpkin
|
| 855 |
+
855:puncher
|
| 856 |
+
856:puppet,marionette
|
| 857 |
+
857:puppy
|
| 858 |
+
858:quesadilla
|
| 859 |
+
859:quiche
|
| 860 |
+
860:quilt,comforter
|
| 861 |
+
861:rabbit
|
| 862 |
+
862:race car,racing car
|
| 863 |
+
863:racket,racquet
|
| 864 |
+
864:radar
|
| 865 |
+
865:radiator
|
| 866 |
+
866:radio receiver,radio set,radio,tuner (radio)
|
| 867 |
+
867:radish,daikon
|
| 868 |
+
868:raft
|
| 869 |
+
869:rag doll
|
| 870 |
+
870:raincoat,waterproof jacket
|
| 871 |
+
871:ram (animal)
|
| 872 |
+
872:raspberry
|
| 873 |
+
873:rat
|
| 874 |
+
874:razorblade
|
| 875 |
+
875:reamer (juicer),juicer,juice reamer
|
| 876 |
+
876:rearview mirror
|
| 877 |
+
877:receipt
|
| 878 |
+
878:recliner,reclining chair,lounger (chair)
|
| 879 |
+
879:record player,phonograph (record player),turntable
|
| 880 |
+
880:reflector
|
| 881 |
+
881:remote control
|
| 882 |
+
882:rhinoceros
|
| 883 |
+
883:rib (food)
|
| 884 |
+
884:rifle
|
| 885 |
+
885:ring
|
| 886 |
+
886:river boat
|
| 887 |
+
887:road map
|
| 888 |
+
888:robe
|
| 889 |
+
889:rocking chair
|
| 890 |
+
890:rodent
|
| 891 |
+
891:roller skate
|
| 892 |
+
892:Rollerblade
|
| 893 |
+
893:rolling pin
|
| 894 |
+
894:root beer
|
| 895 |
+
895:router (computer equipment)
|
| 896 |
+
896:rubber band,elastic band
|
| 897 |
+
897:runner (carpet)
|
| 898 |
+
898:plastic bag,paper bag
|
| 899 |
+
899:saddle (on an animal)
|
| 900 |
+
900:saddle blanket,saddlecloth,horse blanket
|
| 901 |
+
901:saddlebag
|
| 902 |
+
902:safety pin
|
| 903 |
+
903:sail
|
| 904 |
+
904:salad
|
| 905 |
+
905:salad plate,salad bowl
|
| 906 |
+
906:salami
|
| 907 |
+
907:salmon (fish)
|
| 908 |
+
908:salmon (food)
|
| 909 |
+
909:salsa
|
| 910 |
+
910:saltshaker
|
| 911 |
+
911:sandal (type of shoe)
|
| 912 |
+
912:sandwich
|
| 913 |
+
913:satchel
|
| 914 |
+
914:saucepan
|
| 915 |
+
915:saucer
|
| 916 |
+
916:sausage
|
| 917 |
+
917:sawhorse,sawbuck
|
| 918 |
+
918:saxophone
|
| 919 |
+
919:scale (measuring instrument)
|
| 920 |
+
920:scarecrow,strawman
|
| 921 |
+
921:scarf
|
| 922 |
+
922:school bus
|
| 923 |
+
923:scissors
|
| 924 |
+
924:scoreboard
|
| 925 |
+
925:scraper
|
| 926 |
+
926:screwdriver
|
| 927 |
+
927:scrubbing brush
|
| 928 |
+
928:sculpture
|
| 929 |
+
929:seabird,seafowl
|
| 930 |
+
930:seahorse
|
| 931 |
+
931:seaplane,hydroplane
|
| 932 |
+
932:seashell
|
| 933 |
+
933:sewing machine
|
| 934 |
+
934:shaker
|
| 935 |
+
935:shampoo
|
| 936 |
+
936:shark
|
| 937 |
+
937:sharpener
|
| 938 |
+
938:Sharpie
|
| 939 |
+
939:shaver (electric),electric shaver,electric razor
|
| 940 |
+
940:shaving cream,shaving soap
|
| 941 |
+
941:shawl
|
| 942 |
+
942:shears
|
| 943 |
+
943:sheep
|
| 944 |
+
944:shepherd dog,sheepdog
|
| 945 |
+
945:sherbert,sherbet
|
| 946 |
+
946:shield
|
| 947 |
+
947:shirt
|
| 948 |
+
948:shoe,sneaker (type of shoe),tennis shoe
|
| 949 |
+
949:shopping bag
|
| 950 |
+
950:shopping cart
|
| 951 |
+
951:short pants,shorts (clothing),trunks (clothing)
|
| 952 |
+
952:shot glass
|
| 953 |
+
953:shoulder bag
|
| 954 |
+
954:shovel
|
| 955 |
+
955:shower head
|
| 956 |
+
956:shower cap
|
| 957 |
+
957:shower curtain
|
| 958 |
+
958:shredder (for paper)
|
| 959 |
+
959:signboard
|
| 960 |
+
960:silo
|
| 961 |
+
961:sink
|
| 962 |
+
962:skateboard
|
| 963 |
+
963:skewer
|
| 964 |
+
964:ski
|
| 965 |
+
965:ski boot
|
| 966 |
+
966:ski parka,ski jacket
|
| 967 |
+
967:ski pole
|
| 968 |
+
968:skirt
|
| 969 |
+
969:skullcap
|
| 970 |
+
970:sled,sledge,sleigh
|
| 971 |
+
971:sleeping bag
|
| 972 |
+
972:sling (bandage),triangular bandage
|
| 973 |
+
973:slipper (footwear),carpet slipper (footwear)
|
| 974 |
+
974:smoothie
|
| 975 |
+
975:snake,serpent
|
| 976 |
+
976:snowboard
|
| 977 |
+
977:snowman
|
| 978 |
+
978:snowmobile
|
| 979 |
+
979:soap
|
| 980 |
+
980:soccer ball
|
| 981 |
+
981:sock
|
| 982 |
+
982:sofa,couch,lounge
|
| 983 |
+
983:softball
|
| 984 |
+
984:solar array,solar battery,solar panel
|
| 985 |
+
985:sombrero
|
| 986 |
+
986:soup
|
| 987 |
+
987:soup bowl
|
| 988 |
+
988:soupspoon
|
| 989 |
+
989:sour cream,soured cream
|
| 990 |
+
990:soya milk,soybean milk,soymilk
|
| 991 |
+
991:space shuttle
|
| 992 |
+
992:sparkler (fireworks)
|
| 993 |
+
993:spatula
|
| 994 |
+
994:spear,lance
|
| 995 |
+
995:spectacles,specs,eyeglasses,glasses
|
| 996 |
+
996:spice rack
|
| 997 |
+
997:spider
|
| 998 |
+
998:crawfish,crayfish
|
| 999 |
+
999:sponge
|
| 1000 |
+
1000:spoon
|
| 1001 |
+
1001:sportswear,athletic wear,activewear
|
| 1002 |
+
1002:spotlight
|
| 1003 |
+
1003:squid (food),calamari,calamary
|
| 1004 |
+
1004:squirrel
|
| 1005 |
+
1005:stagecoach
|
| 1006 |
+
1006:stapler (stapling machine)
|
| 1007 |
+
1007:starfish,sea star
|
| 1008 |
+
1008:statue (sculpture)
|
| 1009 |
+
1009:steak (food)
|
| 1010 |
+
1010:steak knife
|
| 1011 |
+
1011:steering wheel
|
| 1012 |
+
1012:stepladder
|
| 1013 |
+
1013:step stool
|
| 1014 |
+
1014:stereo (sound system)
|
| 1015 |
+
1015:stew
|
| 1016 |
+
1016:stirrer
|
| 1017 |
+
1017:stirrup
|
| 1018 |
+
1018:stool
|
| 1019 |
+
1019:stop sign
|
| 1020 |
+
1020:brake light
|
| 1021 |
+
1021:stove,kitchen stove,range (kitchen appliance),kitchen range,cooking stove
|
| 1022 |
+
1022:strainer
|
| 1023 |
+
1023:strap
|
| 1024 |
+
1024:straw (for drinking),drinking straw
|
| 1025 |
+
1025:strawberry
|
| 1026 |
+
1026:street sign
|
| 1027 |
+
1027:streetlight,street lamp
|
| 1028 |
+
1028:string cheese
|
| 1029 |
+
1029:stylus
|
| 1030 |
+
1030:subwoofer
|
| 1031 |
+
1031:sugar bowl
|
| 1032 |
+
1032:sugarcane (plant)
|
| 1033 |
+
1033:suit (clothing)
|
| 1034 |
+
1034:sunflower
|
| 1035 |
+
1035:sunglasses
|
| 1036 |
+
1036:sunhat
|
| 1037 |
+
1037:surfboard
|
| 1038 |
+
1038:sushi
|
| 1039 |
+
1039:mop
|
| 1040 |
+
1040:sweat pants
|
| 1041 |
+
1041:sweatband
|
| 1042 |
+
1042:sweater
|
| 1043 |
+
1043:sweatshirt
|
| 1044 |
+
1044:sweet potato
|
| 1045 |
+
1045:swimsuit,swimwear,bathing suit,swimming costume,bathing costume,swimming trunks,bathing trunks
|
| 1046 |
+
1046:sword
|
| 1047 |
+
1047:syringe
|
| 1048 |
+
1048:Tabasco sauce
|
| 1049 |
+
1049:table-tennis table,ping-pong table
|
| 1050 |
+
1050:table
|
| 1051 |
+
1051:table lamp
|
| 1052 |
+
1052:tablecloth
|
| 1053 |
+
1053:tachometer
|
| 1054 |
+
1054:taco
|
| 1055 |
+
1055:tag
|
| 1056 |
+
1056:taillight,rear light
|
| 1057 |
+
1057:tambourine
|
| 1058 |
+
1058:army tank,armored combat vehicle,armoured combat vehicle
|
| 1059 |
+
1059:tank (storage vessel),storage tank
|
| 1060 |
+
1060:tank top (clothing)
|
| 1061 |
+
1061:tape (sticky cloth or paper)
|
| 1062 |
+
1062:tape measure,measuring tape
|
| 1063 |
+
1063:tapestry
|
| 1064 |
+
1064:tarp
|
| 1065 |
+
1065:tartan,plaid
|
| 1066 |
+
1066:tassel
|
| 1067 |
+
1067:tea bag
|
| 1068 |
+
1068:teacup
|
| 1069 |
+
1069:teakettle
|
| 1070 |
+
1070:teapot
|
| 1071 |
+
1071:teddy bear
|
| 1072 |
+
1072:telephone,phone,telephone set
|
| 1073 |
+
1073:telephone booth,phone booth,call box,telephone box,telephone kiosk
|
| 1074 |
+
1074:telephone pole,telegraph pole,telegraph post
|
| 1075 |
+
1075:telephoto lens,zoom lens
|
| 1076 |
+
1076:television camera,tv camera
|
| 1077 |
+
1077:television set,tv,tv set
|
| 1078 |
+
1078:tennis ball
|
| 1079 |
+
1079:tennis racket
|
| 1080 |
+
1080:tequila
|
| 1081 |
+
1081:thermometer
|
| 1082 |
+
1082:thermos bottle
|
| 1083 |
+
1083:thermostat
|
| 1084 |
+
1084:thimble
|
| 1085 |
+
1085:thread,yarn
|
| 1086 |
+
1086:thumbtack,drawing pin,pushpin
|
| 1087 |
+
1087:tiara
|
| 1088 |
+
1088:tiger
|
| 1089 |
+
1089:tights (clothing),leotards
|
| 1090 |
+
1090:timer,stopwatch
|
| 1091 |
+
1091:tinfoil
|
| 1092 |
+
1092:tinsel
|
| 1093 |
+
1093:tissue paper
|
| 1094 |
+
1094:toast (food)
|
| 1095 |
+
1095:toaster
|
| 1096 |
+
1096:toaster oven
|
| 1097 |
+
1097:toilet
|
| 1098 |
+
1098:toilet tissue,toilet paper,bathroom tissue
|
| 1099 |
+
1099:tomato
|
| 1100 |
+
1100:tongs
|
| 1101 |
+
1101:toolbox
|
| 1102 |
+
1102:toothbrush
|
| 1103 |
+
1103:toothpaste
|
| 1104 |
+
1104:toothpick
|
| 1105 |
+
1105:cover
|
| 1106 |
+
1106:tortilla
|
| 1107 |
+
1107:tow truck
|
| 1108 |
+
1108:towel
|
| 1109 |
+
1109:towel rack,towel rail,towel bar
|
| 1110 |
+
1110:toy
|
| 1111 |
+
1111:tractor (farm equipment)
|
| 1112 |
+
1112:traffic light
|
| 1113 |
+
1113:dirt bike
|
| 1114 |
+
1114:trailer truck,tractor trailer,trucking rig,articulated lorry,semi truck
|
| 1115 |
+
1115:train (railroad vehicle),railroad train
|
| 1116 |
+
1116:trampoline
|
| 1117 |
+
1117:tray
|
| 1118 |
+
1118:trench coat
|
| 1119 |
+
1119:triangle (musical instrument)
|
| 1120 |
+
1120:tricycle
|
| 1121 |
+
1121:tripod
|
| 1122 |
+
1122:trousers,pants (clothing)
|
| 1123 |
+
1123:truck
|
| 1124 |
+
1124:truffle (chocolate),chocolate truffle
|
| 1125 |
+
1125:trunk
|
| 1126 |
+
1126:vat
|
| 1127 |
+
1127:turban
|
| 1128 |
+
1128:turkey (food)
|
| 1129 |
+
1129:turnip
|
| 1130 |
+
1130:turtle
|
| 1131 |
+
1131:turtleneck (clothing),polo-neck
|
| 1132 |
+
1132:typewriter
|
| 1133 |
+
1133:umbrella
|
| 1134 |
+
1134:underwear,underclothes,underclothing,underpants
|
| 1135 |
+
1135:unicycle
|
| 1136 |
+
1136:urinal
|
| 1137 |
+
1137:urn
|
| 1138 |
+
1138:vacuum cleaner
|
| 1139 |
+
1139:vase
|
| 1140 |
+
1140:vending machine
|
| 1141 |
+
1141:vent,blowhole,air vent
|
| 1142 |
+
1142:vest,waistcoat
|
| 1143 |
+
1143:videotape
|
| 1144 |
+
1144:vinegar
|
| 1145 |
+
1145:violin,fiddle
|
| 1146 |
+
1146:vodka
|
| 1147 |
+
1147:volleyball
|
| 1148 |
+
1148:vulture
|
| 1149 |
+
1149:waffle
|
| 1150 |
+
1150:waffle iron
|
| 1151 |
+
1151:wagon
|
| 1152 |
+
1152:wagon wheel
|
| 1153 |
+
1153:walking stick
|
| 1154 |
+
1154:wall clock
|
| 1155 |
+
1155:wall socket,wall plug,electric outlet,electrical outlet,outlet,electric receptacle
|
| 1156 |
+
1156:wallet,billfold
|
| 1157 |
+
1157:walrus
|
| 1158 |
+
1158:wardrobe
|
| 1159 |
+
1159:washbasin,basin (for washing),washbowl,washstand,handbasin
|
| 1160 |
+
1160:automatic washer,washing machine
|
| 1161 |
+
1161:watch,wristwatch
|
| 1162 |
+
1162:water bottle
|
| 1163 |
+
1163:water cooler
|
| 1164 |
+
1164:water faucet,water tap,tap (water faucet)
|
| 1165 |
+
1165:water heater,hot-water heater
|
| 1166 |
+
1166:water jug
|
| 1167 |
+
1167:water gun,squirt gun
|
| 1168 |
+
1168:water scooter,sea scooter,jet ski
|
| 1169 |
+
1169:water ski
|
| 1170 |
+
1170:water tower
|
| 1171 |
+
1171:watering can
|
| 1172 |
+
1172:watermelon
|
| 1173 |
+
1173:weathervane,vane (weathervane),wind vane
|
| 1174 |
+
1174:webcam
|
| 1175 |
+
1175:wedding cake,bridecake
|
| 1176 |
+
1176:wedding ring,wedding band
|
| 1177 |
+
1177:wet suit
|
| 1178 |
+
1178:wheel
|
| 1179 |
+
1179:wheelchair
|
| 1180 |
+
1180:whipped cream
|
| 1181 |
+
1181:whistle
|
| 1182 |
+
1182:wig
|
| 1183 |
+
1183:wind chime
|
| 1184 |
+
1184:windmill
|
| 1185 |
+
1185:window box (for plants)
|
| 1186 |
+
1186:windshield wiper,windscreen wiper,wiper (for windshield/screen)
|
| 1187 |
+
1187:windsock,air sock,air-sleeve,wind sleeve,wind cone
|
| 1188 |
+
1188:wine bottle
|
| 1189 |
+
1189:wine bucket,wine cooler
|
| 1190 |
+
1190:wineglass
|
| 1191 |
+
1191:blinder (for horses)
|
| 1192 |
+
1192:wok
|
| 1193 |
+
1193:wolf
|
| 1194 |
+
1194:wooden spoon
|
| 1195 |
+
1195:wreath
|
| 1196 |
+
1196:wrench,spanner
|
| 1197 |
+
1197:wristband
|
| 1198 |
+
1198:wristlet,wrist band
|
| 1199 |
+
1199:yacht
|
| 1200 |
+
1200:yogurt,yoghurt,yoghourt
|
| 1201 |
+
1201:yoke (animal equipment)
|
| 1202 |
+
1202:zebra
|
| 1203 |
+
1203:zucchini,courgette
|