metadata

license: agpl-3.0
language:
  - en
metrics:
  - mae
  - mse
  - accuracy
tags:
  - biology
  - plant
  - vitis
  - downey mildew
  - Plasmopara viticola
  - OIV 452-1
base_model: microsoft/swin-tiny-patch4-window7-224

OIV Leaf Disc Phenotyping

Companion repository for the article "Phenotyping grapevine resistance to downy mildew: deep learning as a promising tool to assess sporulation and necrosis" found Here

Folder Structure

checkpoints

Contains checkpoint files for leaf disc detector and OIV 452-1 scorer.

data

Contains all datasets data in CSV format. All files are semicolon separated.

Leaf Disc Detection Files

ldd_train.csv, ldd_val.csv and ldd_test.csv contain bounding box annotations in Pascal VOC format.
train_ld_bounding_boxes.csv contains predictions for all available plates.

OIV 452-1 Annotation

oiv_annotation.csv fully annotated dataset created with the UI in leaf_patch_annotation.ipynb.
oiv_annotation_empty.csv empty annotation CSV, use it to familiarize yoursel with the annotation process.

OIV 452-1 Predictions

oiv_train.csv, oiv_val.csv and oiv_test.csv contain OIV 452-1 annotated scores.

Genotype Differenciation

genotype_differenciation_dataset.csv contains annotated scores and predictions for leaf patches used in to validate model on genptype differenciation.

images

Contains all images in three different folders:

plates contains full plate images.
leaf_discs contains full leaf discs. Output folder for predicted leaf discs.
leaf_patches contains extracted patches. Output folder for predicted leaf patches.

src

Contains source code under two formats:

*.py files contain base functionality and classes.
*.ipynb files contain code to reproduce the article data.

Notebooks

repo_manager.ipynb

Utility notebook to create this repository