preprocess images and generate train, validation, test dataset this is my first repository
1. Data.py
generate a batch of (images, labels) data.
generate a batch of (images, label1s, label2s) data.
normalize datapoint to unit length (L2 norm equals 1)