AndersonYong/URetinex-Net

How to train it?

Opened this issue · 2 comments

Yukun-Yang-98 commented

Hi! I find that your work performs quite well. Can you tell me how to train your network or provide some files for training?

AndersonYong commented

Hi, the overall network is not trained end-to-end. Three modules are trained speraetly. Loss function for each module are well described in the paper. I am sorry the training is not available for now, please keep following and I will release it later.

  • 👀2
Yukun-Yang-98 commented

Thanks for answering! But I still have a question after reading your paper, one of effects in your unfolding optiminization module is to denoise R, I can't understand how it works. From eq(16), I only find that minimize the gradient of L can help smooth.