I want to extract the features from various CNNs of some images of black silhouettes.
There is important information in the whole picture (due to shape curvatures close to the picture edge).
So, I believe that cropping in this case will throw away this information.
Is it ok if I remove it and only use resizing as pre-processing.?
(I know I also lose some information with resizing, but it's not important in my case).