This repository contains an op-for-op PyTorch reimplementation of Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network.
Contains DIV2K, DIV8K, Flickr2K, OST, T91, Set5, Set14, BSDS100 and BSDS200, etc.
Please refer to README.md in the data directory for the method of making a dataset.
Modify the contents of the config.py file as follows.
- line 29:
upscale_factorchange to the magnification you need to enlarge. - line 31:
modechange Set to valid mode. - line 100:
model_pathchange weight address after training.
Modify the contents of the config.py file as follows.
- line 29:
upscale_factorchange to the magnification you need to enlarge. - line 31:
modechange Set to train mode.
If you want to load weights that you've trained before, modify the contents of the file as follows.
Modify the contents of the config.py file as follows.
- line 47:
start_epochchange number of SRResNet training iterations in the previous round. - line 48:
resumechange to SRResNet weight address that needs to be loaded.
Modify the contents of the config.py file as follows.
- line 71:
start_epochchange number of SRGAN training iterations in the previous round. - line 72:
resumechange to RRDBNet weight address that needs to be loaded. - line 73:
resume_dchange to Discriminator weight address that needs to be loaded. - line 74:
resume_gchange to Generator weight address that needs to be loaded.
Source of original paper results: https://arxiv.org/pdf/1609.04802v5.pdf
In the following table, the psnr value in () indicates the result of the project, and - indicates no test.
| Dataset | Scale | SRResNet (PSNR) | SRGAN (PSNR) |
|---|---|---|---|
| Set5 | 4 | 32.05(32.00) | 29.40(28.91) |
| Set14 | 4 | 28.49(28.28) | 26.02(25.70) |
Low resolution / Recovered High Resolution / Ground Truth
If you find a bug, create a GitHub issue, or even better, submit a pull request. Similarly, if you have questions, simply post them as GitHub issues.
I look forward to seeing what the community does with these models!
Christian Ledig, Lucas Theis, Ferenc Huszar, Jose Caballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Zehan
Wang, Wenzhe Shi
Abstract
Despite the breakthroughs in accuracy and speed of single image super-resolution using faster and deeper convolutional neural networks, one central
problem remains largely unsolved: how do we recover the finer texture details when we super-resolve at large upscaling factors? The behavior of
optimization-based super-resolution methods is principally driven by the choice of the objective function. Recent work has largely focused on
minimizing the mean squared reconstruction error. The resulting estimates have high peak signal-to-noise ratios, but they are often lacking
high-frequency details and are perceptually unsatisfying in the sense that they fail to match the fidelity expected at the higher resolution. In this
paper, we present SRGAN, a generative adversarial network (GAN) for image super-resolution (SR). To our knowledge, it is the first framework capable
of inferring photo-realistic natural images for 4x upscaling factors. To achieve this, we propose a perceptual loss function which consists of an
adversarial loss and a content loss. The adversarial loss pushes our solution to the natural image manifold using a discriminator network that is
trained to differentiate between the super-resolved images and original photo-realistic images. In addition, we use a content loss motivated by
perceptual similarity instead of similarity in pixel space. Our deep residual network is able to recover photo-realistic textures from heavily
downsampled images on public benchmarks. An extensive mean-opinion-score (MOS) test shows hugely significant gains in perceptual quality using SRGAN.
The MOS scores obtained with SRGAN are closer to those of the original high-resolution images than to those obtained with any state-of-the-art method.
@InProceedings{srgan,
author = {Christian Ledig, Lucas Theis, Ferenc Huszar, Jose Caballero, Andrew Cunningham, Alejandro Acosta, Andrew Aitken, Alykhan Tejani, Johannes Totz, Zehan Wang, Wenzhe Shi},
title = {Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network},
booktitle = {arXiv},
year = {2016}
}