Papers › SwinFIR: Revisiting the SwinIR with Fast Fourier Convolution and Improved Training for...

SwinFIR: Revisiting the SwinIR with Fast Fourier Convolution and Improved Training for Image Super-Resolution

24 Aug 2022arXiv:2208.11247archive 2025-07-28

Dafeng Zhang, Feiyu Huang, Shizhuo Liu, Xiaobing Wang, Zhezhu Jin

Transformer-based methods have achieved impressive image restoration performance due to their capacities to model long-range dependency compared to CNN-based methods. However, advances like SwinIR adopts the window-based and local attention strategy to balance the performance and computational overhead, which restricts employing large receptive fields to capture global information and establish long dependencies in the early layers. To further improve the efficiency of capturing global information, in this work, we propose SwinFIR to extend SwinIR by replacing Fast Fourier Convolution (FFC) components, which have the image-wide receptive field. We also revisit other advanced techniques, i.e, data augmentation, pre-training, and feature ensemble to improve the effect of image reconstruction. And our feature ensemble method enables the performance of the model to be considerably enhanced without increasing the training and testing time. We applied our algorithm on multiple popular large-scale benchmarks and achieved state-of-the-art performance comparing to the existing methods. For example, our SwinFIR achieves the PSNR of 32.83 dB on Manga109 dataset, which is 0.8 dB higher than the state-of-the-art SwinIR method.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

IMPLabUniPr/swin2-mose mentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Data AugmentationImage ReconstructionImage RestorationImage Super-ResolutionStereo Image Super-ResolutionSuper-Resolution

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Super-Resolution BSD100 - 2x upscaling HAT_FIR PSNR 32.71 #8 of 30 Archive leaderboard report
Image Super-Resolution BSD100 - 2x upscaling SwinFIR PSNR 32.64 #12 of 30 Archive leaderboard report
Image Super-Resolution BSD100 - 2x upscaling SwinFIR SSIM 0.9054 #12 of 30 Archive leaderboard report
Image Super-Resolution BSD100 - 3x upscaling HAT_FIR PSNR 29.6 #4 of 21 Archive leaderboard report
Image Super-Resolution BSD100 - 3x upscaling SwinFIR PSNR 29.55 #8 of 21 Archive leaderboard report
Image Super-Resolution BSD100 - 3x upscaling SwinFIR SSIM 0.8169 #8 of 21 Archive leaderboard report
Image Super-Resolution BSD100 - 4x upscaling HAT_FIR PSNR 28.07 #5 of 71 Archive leaderboard report
Image Super-Resolution BSD100 - 4x upscaling SwinFIR PSNR 28.03 #10 of 71 Archive leaderboard report
Image Super-Resolution BSD100 - 4x upscaling SwinFIR SSIM 0.7520 #10 of 71 Archive leaderboard report
Image Super-Resolution Manga109 - 2x upscaling HAT_FIR PSNR 40.77 #5 of 21 Archive leaderboard report
Image Super-Resolution Manga109 - 2x upscaling SwinFIR PSNR 40.61 #7 of 21 Archive leaderboard report
Image Super-Resolution Manga109 - 2x upscaling SwinFIR SSIM 0.9816 #7 of 21 Archive leaderboard report
Image Super-Resolution Manga109 - 3x upscaling HAT_FIR PSNR 35.92 #4 of 17 Archive leaderboard report
Image Super-Resolution Manga109 - 3x upscaling SwinFIR PSNR 35.77 #6 of 17 Archive leaderboard report
Image Super-Resolution Manga109 - 3x upscaling SwinFIR SSIM 0.9563 #6 of 17 Archive leaderboard report
Image Super-Resolution Manga109 - 4x upscaling SwinFIR PSNR 32.83 #8 of 50 Archive leaderboard report
Image Super-Resolution Manga109 - 4x upscaling SwinFIR SSIM 0.9314 #8 of 50 Archive leaderboard report
Image Super-Resolution Manga109 - 4x upscaling HAT_FIR PSNR 33.03 #48 of 50 Archive leaderboard report
Image Super-Resolution Set14 - 2x upscaling HAT_FIR PSNR 35.17 #5 of 35 Archive leaderboard report
Image Super-Resolution Set14 - 2x upscaling SwinFIR PSNR 34.93 #9 of 35 Archive leaderboard report
Image Super-Resolution Set14 - 2x upscaling SwinFIR SSIM 0.9276 #9 of 35 Archive leaderboard report
Image Super-Resolution Set14 - 3x upscaling HAT_FIR PSNR 31.37 #4 of 24 Archive leaderboard report
Image Super-Resolution Set14 - 3x upscaling SwinFIR PSNR 31.24 #6 of 24 Archive leaderboard report
Image Super-Resolution Set14 - 3x upscaling SwinFIR SSIM 0.8566 #6 of 24 Archive leaderboard report
Image Super-Resolution Set14 - 4x upscaling HAT_FIR PSNR 29.44 #5 of 104 Archive leaderboard report
Image Super-Resolution Set14 - 4x upscaling SwinFIR PSNR 29.36 #9 of 104 Archive leaderboard report
Image Super-Resolution Set14 - 4x upscaling SwinFIR SSIM 0.7993 #9 of 104 Archive leaderboard report
Image Super-Resolution Set5 - 2x upscaling HAT_FIR PSNR 38.74 #5 of 41 Archive leaderboard report
Image Super-Resolution Set5 - 2x upscaling SwinFIR PSNR 38.65 #10 of 41 Archive leaderboard report
Image Super-Resolution Set5 - 2x upscaling SwinFIR SSIM 0.9633 #10 of 41 Archive leaderboard report
Image Super-Resolution Set5 - 3x upscaling HAT_FIR PSNR 35.21 #4 of 32 Archive leaderboard report
Image Super-Resolution Set5 - 3x upscaling SwinFIR PSNR 35.15 #10 of 32 Archive leaderboard report
Image Super-Resolution Set5 - 3x upscaling SwinFIR SSIM 0.9330 #10 of 32 Archive leaderboard report
Image Super-Resolution Urban100 - 2x upscaling HAT_FIR PSNR 34.94 #5 of 29 Archive leaderboard report
Image Super-Resolution Urban100 - 2x upscaling SwinFIR PSNR 34.57 #9 of 29 Archive leaderboard report
Image Super-Resolution Urban100 - 2x upscaling SwinFIR SSIM 0.9473 #9 of 29 Archive leaderboard report
Image Super-Resolution Urban100 - 3x upscaling HAT_FIR PSNR 30.77 #4 of 22 Archive leaderboard report
Image Super-Resolution Urban100 - 3x upscaling SwinFIR PSNR 30.43 #8 of 22 Archive leaderboard report
Image Super-Resolution Urban100 - 3x upscaling SwinFIR SSIM 0.8913 #8 of 22 Archive leaderboard report
Image Super-Resolution Urban100 - 4x upscaling HAT_FIR PSNR 28.43 #5 of 65 Archive leaderboard report
Image Super-Resolution Urban100 - 4x upscaling SwinFIR PSNR 28.12 #12 of 65 Archive leaderboard report
Image Super-Resolution Urban100 - 4x upscaling SwinFIR SSIM 0.8393 #12 of 65 Archive leaderboard report
Stereo Image Super-Resolution Flickr1024 - 2x upscaling SwinFIRSSR PSNR 30.14 #2 of 10 Archive leaderboard report
Stereo Image Super-Resolution Flickr1024 - 4x upscaling SwinFIRSSR PSNR 24.29 #1 of 8 Archive leaderboard report
Stereo Image Super-Resolution KITTI2012 - 2x upscaling SwinFIRSSR PSNR 31.79 #1 of 5 Archive leaderboard report
Stereo Image Super-Resolution KITTI2012 - 2x upscaling SwinFIRSSR PSNR 31.79 #2 of 5 Archive leaderboard report
Stereo Image Super-Resolution KITTI2012 - 4x upscaling SwinFIRSSR PSNR 27.16 #1 of 9 Archive leaderboard report
Stereo Image Super-Resolution KITTI2015 - 4x upscaling SwinFIRSSR PSNR 26.89 #3 of 9 Archive leaderboard report
Stereo Image Super-Resolution Middlebury - 4x upscaling SwinFIRSSR PSNR 30.44 #1 of 9 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Convolution

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections