SwinIR实战：详细记录SwinIR的训练过程

## Training and testing codes for USRNet, DnCNN, FFDNet, SRMD, DPSR, MSRResNet, ESRGAN, BSRGAN, SwinIR, VRT [](https://github.com/cszn/KAIR/releases)  [Kai Zhang](https://cszn.github.io/) *[Computer Vision Lab](https://vision.ee.ethz.ch/the-institute.html), ETH Zurich, Switzerland* _______ - **_News (2022-03-23)_**: We release [the testing codes](https://github.com/cszn/SCUNet) of [SCUNet ](https://github.com/cszn/SCUNet) for blind real image denoising. __*The following results are obtained by our SCUNet with purely synthetic training data! We did not use the paired noisy/clean data by DND and SIDD during training!*__ <p align="center"> <a href="https://github.com/cszn/SCUNet"> <img width=45% src="https://github.com/cszn/cszn.github.io/blob/master/files/input_16.gif"/> <img width=45% src="https://github.com/cszn/cszn.github.io/blob/master/files/wm_fnb_0010_16.gif"/> </a> </p> - **_News (2022-02-15)_**: We release [the training codes](https://github.com/cszn/KAIR/blob/master/docs/README_VRT.md) of [VRT ](https://github.com/JingyunLiang/VRT) for video SR, deblurring and denoising. <p align="center"> <a href="https://github.com/JingyunLiang/VRT"> <img width=30% src="https://raw.githubusercontent.com/JingyunLiang/VRT/main/assets/teaser_vsr.gif"/> <img width=30% src="https://raw.githubusercontent.com/JingyunLiang/VRT/main/assets/teaser_vdb.gif"/> <img width=30% src="https://raw.githubusercontent.com/JingyunLiang/VRT/main/assets/teaser_vdn.gif"/> </a> </p> - **_News (2021-12-23)_**: Our techniques are adopted in [https://www.amemori.ai/](https://www.amemori.ai/). - **_News (2021-12-23)_**: Our new work for practical image denoising. - <img src="figs/palace.png" height="320px"/> <img src="figs/palace_HSCU.png" height="320px"/> - [<img src="https://github.com/cszn/KAIR/raw/master/figs/denoising_02.png" height="256px"/>](https://imgsli.com/ODczMTc) [<img src="https://github.com/cszn/KAIR/raw/master/figs/denoising_01.png" height="256px"/>](https://imgsli.com/ODczMTY) - **_News (2021-09-09)_**: Add [main_download_pretrained_models.py](https://github.com/cszn/KAIR/blob/master/main_download_pretrained_models.py) to download pre-trained models. - **_News (2021-09-08)_**: Add [matlab code](https://github.com/cszn/KAIR/tree/master/matlab) to zoom local part of an image for the purpose of comparison between different results. - **_News (2021-09-07)_**: We upload [the training code](https://github.com/cszn/KAIR/blob/master/docs/README_SwinIR.md) of [SwinIR ](https://github.com/JingyunLiang/SwinIR) and provide an [interactive online Colob demo for real-world image SR](https://colab.research.google.com/gist/JingyunLiang/a5e3e54bc9ef8d7bf594f6fee8208533/swinir-demo-on-real-world-image-sr.ipynb). Try to super-resolve your own images on Colab! <a href="https://colab.research.google.com/gist/JingyunLiang/a5e3e54bc9ef8d7bf594f6fee8208533/swinir-demo-on-real-world-image-sr.ipynb"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="google colab logo"></a> |Real-World Image (x4)|[BSRGAN, ICCV2021](https://github.com/cszn/BSRGAN)|[Real-ESRGAN](https://github.com/xinntao/Real-ESRGAN)|SwinIR (ours)| | :--- | :---: | :-----: | :-----: | |<img width="200" src="https://raw.githubusercontent.com/JingyunLiang/SwinIR/main/figs/ETH_LR.png">|<img width="200" src="https://raw.githubusercontent.com/JingyunLiang/SwinIR/main/figs/ETH_BSRGAN.png">|<img width="200" src="https://raw.githubusercontent.com/JingyunLiang/SwinIR/main/figs/ETH_realESRGAN.jpg">|<img width="200" src="https://raw.githubusercontent.com/JingyunLiang/SwinIR/main/figs/ETH_SwinIR.png"> |<img width="200" src="https://raw.githubusercontent.com/JingyunLiang/SwinIR/main/figs/OST_009_crop_LR.png">|<img width="200" src="https://raw.githubusercontent.com/JingyunLiang/SwinIR/main/figs/OST_009_crop_BSRGAN.png">|<img width="200" src="https://raw.githubusercontent.com/JingyunLiang/SwinIR/main/figs/OST_009_crop_realESRGAN.png">|<img width="200" src="https://raw.githubusercontent.com/JingyunLiang/SwinIR/main/figs/OST_009_crop_SwinIR.png">| - **_News (2021-08-31)_**: We upload the [training code of BSRGAN](https://github.com/cszn/BSRGAN#training). - **_News (2021-08-24)_**: We upload the BSRGAN degradation model. - **_News (2021-08-22)_**: Support multi-feature-layer VGG perceptual loss and UNet discriminator. - **_News (2021-08-18)_**: We upload the extended BSRGAN degradation model. It is slightly different from our published version. - **_News (2021-06-03)_**: Add testing codes of [GPEN (CVPR21)](https://github.com/yangxy/GPEN) for face image enhancement: [main_test_face_enhancement.py](https://github.com/cszn/KAIR/blob/master/main_test_face_enhancement.py) <img src="figs/face_04_comparison.png" width="730px"/> <img src="figs/face_13_comparison.png" width="730px"/> <img src="figs/face_08_comparison.png" width="730px"/> <img src="figs/face_01_comparison.png" width="730px"/> <img src="figs/face_12_comparison.png" width="730px"/> <img src="figs/face_10_comparison.png" width="730px"/> - **_News (2021-05-13)_**: Add [PatchGAN discriminator](https://github.com/cszn/KAIR/blob/master/models/network_discriminator.py). - **_News (2021-05-12)_**: Support distributed training, see also [https://github.com/xinntao/BasicSR/blob/master/docs/TrainTest.md](https://github.com/xinntao/BasicSR/blob/master/docs/TrainTest.md). - **_News (2021-01)_**: [BSRGAN](https://github.com/cszn/BSRGAN) for blind real image super-resolution will be added. - **_Pull requests are welcome!_** - **Correction (2020-10)**: If you use multiple GPUs for GAN training, remove or comment [Line 105](https://github.com/cszn/KAIR/blob/e52a6944c6a40ba81b88430ffe38fd6517e0449e/models/model_gan.py#L105) to enable `DataParallel` for fast training - **News (2020-10)**: Add [utils_receptivefield.py](https://github.com/cszn/KAIR/blob/master/utils/utils_receptivefield.py) to calculate receptive field. - **News (2020-8)**: A `deep plug-and-play image restoration toolbox` is released at [cszn/DPIR](https://github.com/cszn/DPIR). - **Tips (2020-8)**: Use [this](https://github.com/cszn/KAIR/blob/9fd17abff001ab82a22070f7e442bb5246d2d844/main_challenge_sr.py#L147) to avoid `out of memory` issue. - **News (2020-7)**: Add [main_challenge_sr.py](https://github.com/cszn/KAIR/blob/23b0d0f717980e48fad02513ba14045d57264fe1/main_challenge_sr.py#L90) to get `FLOPs`, `#Params`, `Runtime`, `#Activations`, `#Conv`, and `Max Memory Allocated`. ```python from utils.utils_modelsummary import get_model_activation, get_model_flops input_dim = (3, 256, 256) # set the input dimension activations, num_conv2d = get_model_activation(model, input_dim) logger.info('{:>16s} : {:<.4f} [M]'.format('#Activations', activations/10**6)) logger.info('{:>16s} : {:<d}'.format('#Conv2d', num_conv2d)) flops = get_model_flops(model, input_dim, False) logger.info('{:>16s} : {:<.4f} [G]'.format('FLOPs', flops/10**9)) num_parameters = sum(map(lambda x: x.numel(), model.parameters())) logger.info('{:>16s} : {:<.4f} [M]'.format('#Params', num_parameters/10**6)) ``` - **News (2020-6)**: Add [USRNet (CVPR 2020)](https://github.com/cszn/USRNet) for training and testing. - [Network Architecture](https://github.com/cszn/KAIR/blob/3357aa0e54b81b1e26ceb1cee990f39add235e17/models/network_usrnet.py#L309) - [Dataset](https://github.com/cszn/KAIR/blob/6c852636d3715bb281637863822a42c72739122a/data/dataset_usrnet.py#L16) Clone repo ---------- ``` git clone https://github.com/cszn/KAIR.git ``` ``` pip install -r requirement.txt ```