I saw the code of compare.py and found the function prewhiten.
What is the role of facenet.prewhiten?
It subtracts the average and normalizes the range of the pixel values of input images. It makes training a lot easier.
Is there a method to reverse the prewhiten-ing to reconstruct the original image?
Most helpful comment
It subtracts the average and normalizes the range of the pixel values of input images. It makes training a lot easier.