exactly, that’s the messiness of working with the image format I mentioned. I think you can make to generalize well on all these but since there are more degrees of freedom it will require more data to train (luckily this can be done with automatic data augmentation but still)

→ View original post on X — @rasbt,