KupynOrest / DeblurGAN

Image Deblurring using Generative Adversarial Networks
Other
2.52k stars 521 forks source link

Questions related to ContentLoss design #212

Open xuyu666 opened 3 years ago

xuyu666 commented 3 years ago

Hello author, I am very happy to be able to reproduce your code. I am very curious about your thoughts on design loss, I want to ask some questions about Content_Loss, I hope you can reply.Thanks!!

  1. why did you choose MSE_Loss in your paper?
  2. why do you use the VGG model to extract features as the input of MSE_Loss?
  3. why does VGG happen to be extracted to the 14th layer instead of other quantities?