Awesome-Super-Resolution（in progress）
Collect some super-resolution related papers, data and repositories.
repositories
Awesome paper list:
Awesome repos:
| repo | Framework |
|---|---|
| EDSR-PyTorch | PyTorch |
| Image-Super-Resolution | Keras |
| image-super-resolution | Keras |
| Super-Resolution-Zoo | MxNet |
| super-resolution | Keras |
| neural-enhance | Theano |
| srez | Tensorflow |
| waifu2x | Torch |
| BasicSR | PyTorch |
| super-resolution | PyTorch |
| VideoSuperResolution | Tensorflow |
| video-super-resolution | Pytorch |
Datasets
Note this table is referenced from here.
| Name | Usage | Link | Comments |
|---|---|---|---|
| Set5 | Test | download | jbhuang0604 |
| SET14 | Test | download | jbhuang0604 |
| BSD100 | Test | download | jbhuang0604 |
| Urban100 | Test | download | jbhuang0604 |
| Manga109 | Test | website | |
| SunHay80 | Test | download | jbhuang0604 |
| BSD300 | Train/Val | download | |
| BSD500 | Train/Val | download | |
| 91-Image | Train | download | Yang |
| DIV2K2017 | Train/Val | website | NTIRE2017 |
| Real SR | Train/Val | website | NTIRE2019 |
| Waterloo | Train | website | |
| VID4 | Test | download | 4 videos |
| MCL-V | Train | website | 12 videos |
| GOPRO | Train/Val | website | 33 videos, deblur |
| CelebA | Train | website | Human faces |
| Sintel | Train/Val | website | Optical flow |
| FlyingChairs | Train | website | Optical flow |
| Vimeo-90k | Train/Test | website | 90k HQ videos |
Dataset collections
Benckmark and DIV2K: Set5, Set14, B100, Urban100, Manga109, DIV2K2017 include bicubic downsamples with x2,3,4,8
SR_testing_datasets: Test: Set5, Set14, B100, Urban100, Manga109, Historical; Train: T91,General100, BSDS200
paper
Non-DL based approach
SCSR: TIP2010, Jianchao Yang et al.paper, code
ANR: ICCV2013, Radu Timofte et al. paper, code
A+: ACCV 2014, Radu Timofte et al. paper, code
IA: CVPR2016, Radu Timofte et al. paper
SelfExSR: CVPR2015, Jia-Bin Huang et al. paper, code
NBSRF: ICCV2015, Jordi Salvador et al. paper
RFL: ICCV2015, Samuel Schulter et al paper, code
DL based approach
Note this table is referenced from here
| Model | Published | Code | Keywords |
|---|---|---|---|
| SRCNN | ECCV14 | Keras | Kaiming |
| RAISR | arXiv | - | Google, Pixel 3 |
| ESPCN | CVPR16 | Keras | Real time/SISR/VideoSR |
| VDSR | CVPR16 | Matlab | Deep, Residual |
| DRCN | CVPR16 | Matlab | Recurrent |
| DRRN | CVPR17 | Caffe, PyTorch | Recurrent |
| LapSRN | CVPR17 | Matlab | Huber loss |
| IRCNN | CVPR17 | Matlab | |
| EDSR | CVPR17 | PyTorch | NTIRE17 Champion |
| BTSRN | CVPR17 | - | NTIRE17 |
| SelNet | CVPR17 | - | NTIRE17 |
| TLSR | CVPR17 | - | NTIRE17 |
| SRGAN | CVPR17 | Tensorflow | 1st proposed GAN |
| VESPCN | CVPR17 | - | VideoSR |
| MemNet | ICCV17 | Caffe | |
| SRDenseNet | ICCV17 | -, PyTorch | Dense |
| SPMC | ICCV17 | Tensorflow | VideoSR |
| EnhanceNet | ICCV17 | TensorFlow | Perceptual Loss |
| PRSR | ICCV17 | TensorFlow | an extension of PixelCNN |
| AffGAN | ICLR17 | - | |
| MS-LapSRN | TPAMI18 | Matlab | Fast LapSRN |
| DCSCN | arXiv | Tensorflow | |
| IDN | CVPR18 | Caffe | Fast |
| DSRN | CVPR18 | TensorFlow | Dual state，Recurrent |
| RDN | CVPR18 | Torch | Deep, BI-BD-DN |
| SRMD | CVPR18 | Matlab | Denoise/Deblur/SR |
| DBPN | CVPR18 | PyTorch | NTIRE18 Champion |
| WDSR | CVPR18 | PyTorch，TensorFlow | NTIRE18 Champion |
| ProSRN | CVPR18 | PyTorch | NTIRE18 |
| ZSSR | CVPR18 | Tensorflow | Zero-shot |
| FRVSR | CVPR18 | VideoSR | |
| DUF | CVPR18 | Tensorflow | VideoSR |
| TDAN | arXiv | - | VideoSR，Deformable Align |
| SFTGAN | CVPR18 | PyTorch | |
| CARN | ECCV18 | PyTorch | Lightweight |
| RCAN | ECCV18 | PyTorch | Deep, BI-BD-DN |
| MSRN | ECCV18 | PyTorch | |
| SRFeat | ECCV18 | Tensorflow | GAN |
| ESRGAN | ECCV18 | PyTorch | PRIM18 region 3 Champion |
| FEQE | ECCV18 | Tensorflow | Fast |
| NLRN | NIPS18 | Tensorflow | Non-local, Recurrent |
| SRCliqueNet | NIPS18 | - | Wavelet |
| CBDNet | arXiv | Matlab | Blind-denoise |
| TecoGAN | arXiv | Tensorflow | VideoSR GAN |
| RBPN | CVPR19 | PyTorch | VideoSR |
| SRFBN | CVPR19 | PyTorch | Feedback |
| MoreMNAS | arXiv | - | Lightweight，NAS |
| FALSR | arXiv | TensorFlow | Lightweight，NAS |
| Meta-SR | arXiv | Arbitrary Magnification | |
| AWSRN | arXiv | PyTorch | Lightweight |
| OISR | CVPR19 | PyTorch | ODE-inspired Network |
| DPSR | CVPR19 | PyTorch | |
| DNI | CVPR19 | PyTorch | |
| MAANet | arXiv | Multi-view Aware Attention | |
| RNAN | ICLR19 | PyTorch | Residual Non-local Attention |
| FSTRN | CVPR19 | - | VideoSR, fast spatio-temporal residual block |
| MsDNN | arXiv | TensorFlow | NTIRE19 real SR 21th place |
| SAN | CVPR19 | Pytorch | Second-order Attention,cvpr19 oral |
| EDVR | CVPR19 | Pytorch | Video, NTIRE19 video restoration and enhancement champions |
| Ensemble for VSR | CVPR19 | - | VideoSR, NTIRE19 video SR 2nd place |
| TENet | arXiv | - | a Joint Solution for Demosaicking, Denoising and Super-Resolution |
| MCAN | arXiv | Pytorch | Matrix-in-matrix CAN, Lightweight |
Super Resolution survey：
[1] Wenming Yang, Xuechen Zhang, Yapeng Tian, Wei Wang, Jing-Hao Xue. Deep Learning for Single Image Super-Resolution: A Brief Review. arxiv, 2018. paper
[2]Saeed Anwar, Salman Khan, Nick Barnes. A Deep Journey into Super-resolution: A survey. arxiv, 2019.paper