We have three datasets which are CheXpert, MIMIC-CXR and BRAX. However, we experiment based on just CheXpert. So, I think we have to support MIMIC-CXR and BRAX working well, and experiment them together.
Key Changes 🔑
data_loader/data_loader.py : Remove labeler_path in CXRDataset class, change the way that loads MIMIC dataset, append a way that loads BRAX dataset.
config/Dataset/MIMIC.yaml: Remove labeler_path.
config/Dataset/BRAX.yaml: Add this file and maintain other dataset format.
To Reviewers 🙏
No conflicts.
If you have any questions, feel free to ask.
Support multiple datasets at the same time in the process of train or valid or test will be resolved as soon as possible.
Motivation 🤔
Key Changes 🔑
To Reviewers 🙏
resolves: #85 (Half)