Solid code, it's easy to read and written in a smart way! 馃挴
I've been playing with your code the past days and noticed a mistake in the coloring. It skips the blue and green spectrum of the ab in Lab. It's because of the relu activation in the last layer. It only allows for positive numbers, while the ab space goes from -128 to 128. The below fixes worked for me.
Ytest = Ytest / 128
model.add(Activation('tanh'))
output = output * 128
Solid code, it's easy to read and written in a smart way! 馃挴
I've been playing with your code the past days and noticed a mistake in the coloring. It skips the blue and green spectrum of the ab in Lab. It's because of the relu activation in the last layer. It only allows for positive numbers, while the ab space goes from -128 to 128. The below fixes worked for me.