This is not an issue but curious to know why do we have resizedimages folder. Code worked fine without that folder too.
But if I have to use this code on my own dataset, can I resize it first and run the trainer to save time ? Will the accuracy be same ?