How to run full pipeline

[DanielNajarian]

I wanted to clarify some things about running the code from scratch.

1. Do I need to clone the original nnUNet repo or does this repo work as a standalone?
2. Is there a preprocessing script? I saw some code in the task500 file that looks like it but am unsure if jumping straight to the training would do the proper preprocessing
3. Are the only commands to run the training and the inference?

Thanks

[rixez]

Hi @DanielNajarian,

1. You can run this repo as a standalone. This repo contained all functionalities of the original nnUNet repo.
2. You should first convert your data to the expected format (detailed here). Then, you can run the preprocessing command of nnUNet to automatically prepare your dataset.
3. I include only the sample commands for training and inference. For more commands, you can check out the codes or the original nnUNet repo.

Thanks.

[DanielNajarian]

When running the model, I get an error on the nnunet import statement saying "ModuleNotFoundError: No module named 'nnunet'". Should I pip install nnunet? It wasn't in the requirements.txt and I assumed it wasn't needed since the folder was in the repo.

[rixez]

Please run this command at the root folder to install the repo as a library: pip install -e .