KAIR
KAIR copied to clipboard
USRNet loss
Hello, I am training USRNet for a deblurring task and now it seems to have a nice result after tring some noise level. But I have a quetion when I read the paper 'Deep Unfolding Network for Image Super-Resolution'. It is mentioned that 'we further adopt a weighted combination of L1 loss,VGG perceptual loss and relativistic adversarial loss [24]with weights 1, 1 and 0.005 for perceptual quality performance.', but I couldn't find the relativistic adversarial loss in the code(loss.py) so I only use the l1 and VGG loss. Could you tell me how to add the relativistic adversarial loss? I'm looking forward to your reply