Research map / Image generation / Super-resolution
HAT: Hybrid Attention Transformer for Image Restoration
Xiangyu Chen, Xintao Wang, Wenlong Zhang, Xiangtao Kong, Yu Qiao, Jiantao Zhou, Chao Dong
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| enlarge window size from 8 to 16 (ablation) | SwinIR-style baseline · Urban100 x4 SR | PSNR: 27.45dB -> 27.81dB | Improved |
| add OCAB (ablation) | HAT baseline · Urban100 x4 SR | PSNR: 27.81dB -> 27.91dB | Improved |
| add CAB (ablation) | HAT baseline · Urban100 x4 SR | PSNR: 27.81dB -> 27.91dB | Improved |
| add both OCAB and CAB (ablation) | HAT baseline · Urban100 x4 SR | PSNR: 27.81dB -> 27.97dB | Improved |
| add channel attention in CAB (ablation) | HAT · Urban100 x4 SR | PSNR: 27.92dB -> 27.97dB | Improved |
| use small CAB weight alpha=0.01 instead of 1 (ablation) | HAT · Urban100 x4 SR | PSNR: 27.86dB -> 27.97dB | Improved |
| set overlap ratio gamma 0.5 instead of 0 (ablation) | HAT OCAB · Urban100 x4 SR | PSNR: 27.85dB -> 27.91dB | Improved |
| overlap ratio gamma 0.25 instead of 0 (ablation) | HAT OCAB · Urban100 x4 SR | PSNR: 27.85dB -> 27.81dB | Hurt |
| reduce CAB squeeze factor beta from 6 to 1 (ablation) | HAT · Urban100 x4 SR | PSNR: 27.88dB -> 27.97dB | Improved |
| same-task instead of multi-related-task pre-training (ablation) | HAT · Urban100 x4 SR | PSNR: 28.21dB -> 28.28dB | Improved |
| add ImageNet same-task pre-training | HAT · Urban100 x4 SR | PSNR: 27.97 -> 28.37 | Improved |
| HAT-S versus SwinIR at similar compute | HAT-S · Urban100 x4 SR | PSNR: 27.45dB -> 27.80dB | Improved |
Closest papers by meaning
- Activating More Pixels in Image Super-Resolution Transformer (CVPR 2023 · similarity 0.92)
- Sharing Key Semantics in Transformer Makes Efficient Image Restoration (NeurIPS 2024 · similarity 0.88)
- Dual-Scale Transformer for Large-Scale Single-Pixel Imaging (CVPR 2024 · similarity 0.87)
- Adapt or Perish: Adaptive Sparse Transformer with Attentive Feature Refinement for Image Restoration (CVPR 2024 · similarity 0.87)
- ACL: Activating Capability of Linear Attention for Image Restoration (CVPR 2025 · similarity 0.86)
- Intra and Inter Parser-Prompted Transformers for Effective Image Restoration (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
