I have the problem that while cropping face images (with the same output size 256x256) the face looks bigger in some images than in another ones(the others have more background and less face). I want this images for training a LSTM, so it is this different face's sizes a problem for the training?
Thank you,