MIL-WSI provides APIs to easily train and test state-of-the-art MIL models
conda env create -f enviroment.yaml
Run with one GPU or CPU:
python scripts/extract_features.py --data_path [DATA_PATH] --feature_path [FEATURE_PATH]
Run with multiple GPUs:
torchrun --nproc-per-node [GPU_NUM] scripts/extract_features.py --launcher pytorch --data_path [DATA_PATH] --feature_path [FEATURE_PATH]