Photo-Realistic-Super-Resoluton
Photo-Realistic-Super-Resoluton copied to clipboard
VGG net for content loss
Hi, Harry,
thanks for sharing this code. I have several questions:
-
did you try to add the VGG content loss for the training? It seems that you did not include it in the code?
-
In the original paper, the authors used sub-pixel conv for increasing the resolutions, here you used the spatialfullconv layer in torch. What do you think about the difference, have you tried the sub-pixel conv?
Looking forward to get your feedback!
JIanwei