How to select appropriate transformation for image classification

I don’t know how to apply correct transformations for image classification. Do you have any suggestion for that?