I've just tried to create a new train/test split using the edited labels. It created an iteration-0 folder with CollectedData csv and h5 file and metadata yaml file but DLC is now hanging. Previously I saw it also created a Documentation data pickle file and Matlab file. When I go to train the network, it says "failed to read shuffle 1".

FYI when I first tried to create the train/test split it complained that there were duplicates of the videos in the config file - not sure why, but I edited the pathing to fix it.

Not sure what the issue is - all video folders have a CollectedData .csv and .h5 file now, plus some have machine labels too. Let me know if you'd like me to send over the config file or an example folder.

Originally posted by @d45ha in napari-deeplabcut #189