Now is going to make a low illumination image enhancement of CNN, because is a small white, encounter a problem, data input and label are images, namely a lot of pictures, many are about the classification of CNN do online, so want to ask here, for that matter, using tensorflow how to deal with data sets,

CodePudding user response:

Can see autoencoder architecture, and then selects the L2 loss
Reference: https://en.wikipedia.org/wiki/Autoencoder