/xray-lightning

Chest X-ray object detection with Pytorch Lightning

Primary LanguageJupyter NotebookMIT LicenseMIT

xray-lightning

Pytorch Lightning object detection pipeline for Chest X-ray data

Training

python train.py data.num_workers=10 trainer.max_epochs=3

Inference

python run.py logging.best_model_path=outputs/2020-07-06/22-47-46/best_model/model.pth data.batch_size=1 data.num_workers=1

This will generate predictions.csv that can be used for future analysis.

Note: To do predictions for random 100 images instead of entire dataset, also add testing.debug=True to command line arguments

Results

Some of the results can be found in this notebook.