2019CVPR超分文章网络结构

1、Modulating Image Restoration with Continual Levels via Adaptive Feature Modification Layers

调整模糊度，upscale重建

加入AdaFM模块引导。

2、Natural and Realistic Single Image Super-Resolution with Explicit Natural Manifold Discrimination

3、ODE-inspired Network Design for Single Image Super-Resolution

用残差来体现ODE思想，pixel-shuffle超分

4、Deep Plug-and-Play Super-Resolution for Arbitrary Blur Kernels

使用SRResNet的修改版本，RResNet+，做得修改有
1、增加了一个噪声map作为输入
2、increases the number of fea- ture maps from 64 to 96
3、removes the batch normalization layer

5、Towards Real Scene Super-Resolution with Raw Images

6、Feedback Network for Image Super-Resolution

7、Learning Parallax Attention for Stereo Image Super-Resolution

Stereo image pairs

8、Meta-SR: A Magnification-Arbitrary Network for Super-Resolution

使用RDN作为特征学习模块。

9、Fast Spatio-Temporal Residual Network for Video Super-Resolution

SRResNet

10、Recurrent Back-Projection Network for Video Super-Resolution

11、Blind Super-Resolution With Iterative Kernel Correction