weixin_43840215的博客

CVPR2020所有论文：https://openaccess.thecvf.com/CVPR20201.Unified Dynamic Convolutional Network for Super-Resolution with Variational DegradationsproblemDeep Convolutional Neural Networks (CNNs) have achieved remarkable results on Single Image Super-Resoluti

CVPR2020所有论文：https://openaccess.thecvf.com/CVPR20201.Closed-loop Matters: Dual Regression Networks for Single Image Super-ResolutionproblemTwo underlying limitations to existing SR methods:Learning the mapping function from LR to HR images is typicall

1.Investigating Loss Functions for Extreme Super-ResolutionNTIRE2020极限超分亚军方案，出自CIPLab。LPIPS指标亚军，PI指标冠军。官方的评测指标是LPIPS，所以屈居亚军。以往的超分辨方法主要是做4倍超分，很少一部分工作是16倍超分的。problemThe general approach for perceptual ×4 SR is using GAN with VGG based perceptual loss, ho