https://github.com/mingtaoguo/residual-dense-network-trained-with-cgan-for-super-resolution

This repository is as a research project in the field of super resolution. It uses RDN as the generator and spectral norm is used in discriminator.
https://github.com/mingtaoguo/residual-dense-network-trained-with-cgan-for-super-resolution

cgan rdn superresolution tensorflow

Last synced: 3 months ago
JSON representation

This repository is as a research project in the field of super resolution. It uses RDN as the generator and spectral norm is used in discriminator.

• Host: GitHub
• URL: https://github.com/mingtaoguo/residual-dense-network-trained-with-cgan-for-super-resolution
• Owner: MingtaoGuo
• License: mit
• Created: 2018-12-21T12:25:12.000Z (almost 7 years ago)
• Default Branch: master
• Last Pushed: 2020-06-05T07:53:33.000Z (over 5 years ago)
• Last Synced: 2025-04-05T13:11:18.337Z (7 months ago)
• Topics: cgan, rdn, superresolution, tensorflow
• Language: Python
• Size: 955 KB
• Stars: 25
• Watchers: 1
• Forks: 3
• Open Issues: 2
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution

This repository is as a research project in the field of super resolution. It uses RDN as the generator and spectral norm is used in discriminator.

# Introduction

### This is a trial for super-resolution

The residual dense network has many advantages for reconstructing SR images, and we use GANs to enhance RDN.

The core idea is from the following two papers:

1. Residual Dense Network for Image Super-Resolution

2. cGANs with projection discriminator

##### Generator: Residual Dense Network

![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/RDN.jpg)

##### Discriminator: cGAN projection

![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/pro.jpg)

# Results

These results is just trained about 200,000 iterations (full: 600,000) with batch size of 16. 

|Raw|Bicubic(x4)|RDN_GAN(x4)|

|-|-|-|

|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/hr_flow.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/bic_flow.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/sr_flow.jpg)|

|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/hr_butf.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/bic_butf.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/sr_butf.jpg)|

|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/hr_dog.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/bic_dog.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/sr_dog.jpg)|

|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/hr_cat.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/bic_cat.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/sr_cat.jpg)|

|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/hr_wom.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/bic_wom.jpg)|![](https://github.com/MingtaoGuo/Residual-Dense-Network-Trained-with-cGAN-for-Super-Resolution/blob/master/IMAGES/sr_wom.jpg)|

# Reference 

[1] Zhang Y, Tian Y, Kong Y, et al. Residual dense network for image super-resolution[C]//The IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2018.

[2] Miyato T, Koyama M. cGANs with projection discriminator[J]. arXiv preprint arXiv:1802.05637, 2018.