Repository navigation
Added missing code in prepare_radboud_data_for_training.py to save the proper metadata in metadata.mat. Additionally fixed its behavior that would make the metadata.mat load fail in deepus.py load_header when the most recently saved metadata.mat file is not from one subsampled angle. This behavior has not been patched for the matlab code.
Additionally fixed the default save location of the data prepared by prepare_radboud_data_for_training.py to match what train.py expects.