Papers › HMANet: Hybrid Multi-Axis Aggregation Network for Image Super-Resolution

HMANet: Hybrid Multi-Axis Aggregation Network for Image Super-Resolution

8 May 2024arXiv:2405.05001archive 2025-07-28

Shu-Chuan Chu, Zhi-chao Dou, Jeng-Shyang Pan, Shaowei Weng, JunBao Li

Transformer-based methods have demonstrated excellent performance on super-resolution visual tasks, surpassing conventional convolutional neural networks. However, existing work typically restricts self-attention computation to non-overlapping windows to save computational costs. This means that Transformer-based networks can only use input information from a limited spatial range. Therefore, a novel Hybrid Multi-Axis Aggregation network (HMA) is proposed in this paper to exploit feature potential information better. HMA is constructed by stacking Residual Hybrid Transformer Blocks(RHTB) and Grid Attention Blocks(GAB). On the one side, RHTB combines channel attention and self-attention to enhance non-local feature fusion and produce more attractive visual results. Conversely, GAB is used in cross-domain information interaction to jointly model similar features and obtain a larger perceptual field. For the super-resolution task in the training phase, a novel pre-training method is designed to enhance the model representation capabilities further and validate the proposed model's effectiveness through many experiments. The experimental results show that HMA outperforms the state-of-the-art methods on the benchmark dataset. We provide code and models at https://github.com/korouuuuu/HMA.

PaperPDFCode

Code

korouuuuu/hma officialmentioned in paperpytorch 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

Image Super-Resolution

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Super-Resolution BSD100 - 2x upscaling HMA† PSNR 32.79 #4 of 30 Archive leaderboard report
Image Super-Resolution BSD100 - 2x upscaling HMA† SSIM 0.9071 #4 of 30 Archive leaderboard report
Image Super-Resolution BSD100 - 3x upscaling HMA† PSNR 29.66 #2 of 21 Archive leaderboard report
Image Super-Resolution BSD100 - 3x upscaling HMA† SSIM 0.8196 #2 of 21 Archive leaderboard report
Image Super-Resolution BSD100 - 4x upscaling HMA† PSNR 28.13 #3 of 71 Archive leaderboard report
Image Super-Resolution BSD100 - 4x upscaling HMA† SSIM 0.7562 #3 of 71 Archive leaderboard report
Image Super-Resolution Manga109 - 2x upscaling HMA† PSNR 41.13 #3 of 21 Archive leaderboard report
Image Super-Resolution Manga109 - 2x upscaling HMA† SSIM 0.9836 #3 of 21 Archive leaderboard report
Image Super-Resolution Manga109 - 3x upscaling HMA† PSNR 36.10 #2 of 17 Archive leaderboard report
Image Super-Resolution Manga109 - 3x upscaling HMA† SSIM 0.9580 #2 of 17 Archive leaderboard report
Image Super-Resolution Manga109 - 4x upscaling HMA† PSNR 33.19 #3 of 50 Archive leaderboard report
Image Super-Resolution Manga109 - 4x upscaling HMA† SSIM 0.9344 #3 of 50 Archive leaderboard report
Image Super-Resolution Set14 - 2x upscaling HMA† PSNR 35.33 #2 of 35 Archive leaderboard report
Image Super-Resolution Set14 - 2x upscaling HMA† SSIM 0.9297 #2 of 35 Archive leaderboard report
Image Super-Resolution Set14 - 3x upscaling HMA† PSNR 31.47 #2 of 24 Archive leaderboard report
Image Super-Resolution Set14 - 3x upscaling HMA† SSIM 0.8585 #2 of 24 Archive leaderboard report
Image Super-Resolution Set14 - 4x upscaling HMA† PSNR 29.51 #2 of 104 Archive leaderboard report
Image Super-Resolution Set14 - 4x upscaling HMA† SSIM 0.8019 #2 of 104 Archive leaderboard report
Image Super-Resolution Set5 - 2x upscaling HMA† PSNR 38.95 #2 of 41 Archive leaderboard report
Image Super-Resolution Set5 - 2x upscaling HMA† SSIM 0.9649 #2 of 41 Archive leaderboard report
Image Super-Resolution Set5 - 3x upscaling HMA† PSNR 35.35 #1 of 32 Archive leaderboard report
Image Super-Resolution Set5 - 3x upscaling HMA† SSIM 0.9347 #1 of 32 Archive leaderboard report
Image Super-Resolution Set5 - 4x upscaling HMA† PSNR 33.38 #1 of 12 Archive leaderboard report
Image Super-Resolution Set5 - 4x upscaling HMA† SSIM 0.9089 #1 of 12 Archive leaderboard report
Image Super-Resolution Urban100 - 2x upscaling HMA† PSNR 35.24 #1 of 29 Archive leaderboard report
Image Super-Resolution Urban100 - 2x upscaling HMA† SSIM 0.9513 #1 of 29 Archive leaderboard report
Image Super-Resolution Urban100 - 3x upscaling HMA† PSNR 31.00 #2 of 22 Archive leaderboard report
Image Super-Resolution Urban100 - 3x upscaling HMA† SSIM 0.8984 #2 of 22 Archive leaderboard report
Image Super-Resolution Urban100 - 4x upscaling HMA† PSNR 28.69 #3 of 65 Archive leaderboard report
Image Super-Resolution Urban100 - 4x upscaling HMA† SSIM 0.8512 #3 of 65 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

Absolute Position EncodingsAdamAttentionBPEDense ConnectionsDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxTransformer

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