Computer vision image pre processing

Hi all , any suggestions on what are the ways which we follow in image preparations inter in computer vision

1 Like

1)Try going with grey-scaled images if possible at first for a few of images from your dataset this way you can have a variation in your dataset.
2)Normalization helps as well
3)Since NN are data hungry & then Convolution are just greedy try going with some image augmentation that way you can create more data plus you reduce the bias of your model.
4)In case you’re going with transfer learning then there are special pre-processing functions for each model since on that parameter the models work very best .So try applying that function on your dataset.