Hi and thanks for the contribution
I was wondering why the train and test aren't the same pipeline especially for the dehaze you don't do patch crop and it s forced to read rain test set instead of haze
also how are you reporting the results since the model don't do evaluation for each epoch, is it that later you iterate over all ckpts?
Hi and thanks for the contribution I was wondering why the train and test aren't the same pipeline especially for the dehaze you don't do patch crop and it s forced to read rain test set instead of haze also how are you reporting the results since the model don't do evaluation for each epoch, is it that later you iterate over all ckpts?
Thanks