In order to write a paper outlining our approach in a project it would be really useful to have some transparency as to how many additional image samples are created with the Data Augmentation functionality and what behavior is being applied in terms of combining augmentations.
For example have 250k source image chips, and zoom, contrast, brightness, and rotate selected. I have the following questions regarding what is actually going on in the background:
Is this making 1 copy for each source chip per data augmentation type or is it making several?
Is it one type of augmentation per output chip or are multiple augmentations applied to the same output sample?
Are these created and cached before the model is trained or are they generated dynamically throughout the process?
Is there anyway to see the folder where the augmented image chips are being created?
Many thanks in advance!
