I have just gone through Tutorial 1: Making Your Dataset From Images, and have a couple of basic questions:
1. What is the effect of choosing kernelsz = 7x7? These MNIST images don't seem to be resized, so I would assume that this is somehow tied in with preprocessing
2. Are there preprocessing options? The tutorial seems to imply that there are, but doesn't make their usage explicit.
Thanks.