/covid19_weak_supervision

WACV2021 - A Weakly Supervised Consistency-based Learning Method for COVID-19 Segmentation in CT Images

Primary LanguagePythonApache License 2.0Apache-2.0

A Weakly Supervised Consistency-based Learning Method for COVID-19 Segmentation in CT Images [Paper]

Accepted at WACV2021 Conference. Try the code in this colab.

1. Download Datasets

COVID19

COVID19 V2

COVID19 V3

2. Train & Validate

Run the following command to reproduce the experiments in the paper:

python trainval.py -e weakly_covid19_${DATASET}_${SPLIT} -sb ${SAVEDIR_BASE} -d ${DATADIR} -r 1

The variables (${...}) can be substituted with the following values:

  • DATASET (the COVID dataset): v1, v2, or v3
  • SPLIT (the dataset split): mixed_c2, sep_c2
  • SAVEDIR_BASE: Absolute path to where results will be saved
  • DATADIR: Absolute path containing the downloaded datasets

Experiment hyperparameters are defined in ./exp_configs/weakly_exps.py

3. Visualize the Results

Open results.ipynb for visualization.

Cite

@article{laradji2020weakly,
  title={A Weakly Supervised Consistency-based Learning Method for COVID-19 Segmentation in CT Images},
  author={Laradji, Issam and Rodriguez, Pau and Manas, Oscar and Lensink, Keegan and Law, Marco and Kurzman, Lironne and Parker, William and Vazquez, David and Nowrouzezahrai, Derek},
  journal={arXiv preprint arXiv:2007.02180},
  year={2020}
}