Lung Nodule CT Detection LUNA16 data split

[junxiant]

Hi team,

I was looking at the lung nodule detection training, specifically the data splits. In the tutorial guide, it says:

https://github.com/Project-MONAI/tutorials/tree/main/detection#32-3d-detection-training

"For each fold, 95% of the training data is used for training, while the rest 5% is used for validation and model selection."

And in the training code, it does that exactly, data=train_data[int(0.95 * len(train_data)) :]

However, the downloaded json files from this link https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/LUNA16_datasplit-20220615T233840Z-001.zip

contains both 'training' and 'validation' keys. Since the 5% validation is already split by the code, what will the json file's 'validation' key be used for?

Thanks

[ericspod]

Hi @Can-Zhao @dongyang0122 I think you've worked on this tutorial before, would either have any insights here? Thanks!