Thanks for your extradionary work.
The LR downsampling method of the published training set is Bicubic, the validation set is Gaussian and sampling. Using different method for training will lead to the training of FRVSR and VESPCN stuck when you look at the validation psnr.
Thanks for your extradionary work.
The LR downsampling method of the published training set is Bicubic, the validation set is Gaussian and sampling. Using different method for training will lead to the training of FRVSR and VESPCN stuck when you look at the validation psnr.