Research map / Image generation / Super-resolution
Exploring Frequency-Inspired Optimization in Transformer for Efficient Single Image Super-Resolution
Ao Li, Le Zhang, Yun Liu, Ce Zhu
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 |
|---|---|---|---|
| add SRWAB transformer branch to convolution-only model (ablation) | CRAFT · Manga109 x4 | PSNR: 30.79 -> 31.18 | Improved |
| add HFERB high-frequency branch to transformer-only model (ablation) | CRAFT · Manga109 x4 | PSNR: 31.12 -> 31.18 | Improved |
| use HFB inter-attention fusion instead of concatenation (ablation) | CRAFT · Manga109 x4 | PSNR: 30.92 -> 31.18 | Improved |
| use HFERB as query prior, SRWAB as key/value (vs swapped) (ablation) | CRAFT · Manga109 x4 | PSNR: 30.67 -> 31.18 | Improved |
| use parallel structure instead of cascade (ablation) | CRAFT · Manga109 x4 | PSNR: 30.88 -> 31.18 | Improved |
| remove the attention mask in rectangle attention (ablation) | CRAFT · Manga109 x4 | inference time: 87.44 -> 55.36 | Improved |
| add adaptive dual clipping instead of DBDC (ablation) | PTQ of CRAFT 4-bit · DIV2K x4 | PSNR: 21.36 -> 24.90 | Improved |
| add frequency-guided optimization to ADC (ablation) | PTQ of CRAFT 4-bit · DIV2K x4 | PSNR: 24.90 -> 25.06 | Improved |
| add boundary refinement instead of PaC (ablation) | PTQ of CRAFT 4-bit · Manga109 x4 | PSNR: 25.63 -> 27.07 | Improved |
| add FGO to boundary refinement (ablation) | PTQ of CRAFT 4-bit · Set5 x4 | PSNR: 28.95 -> 29.46 | Improved |
| use frequency-guided PTQ instead of PTQ4SR (ablation) | CRAFT 4-bit · Manga109 x4 | PSNR: 24.81 -> 27.03 | Improved |
| reduce activation/weight precision from 8 to 4 bit | CRAFT with our PTQ · Manga109 x4 | PSNR: 31.06 -> 27.03 | Hurt |
Closest papers by meaning
- AdaFormer: Efficient Transformer with Adaptive Token Sparsification for Image Super-resolution (AAAI 2024 · similarity 0.88)
- Recursive Generalization Transformer for Image Super-Resolution (ICLR 2024 · similarity 0.88)
- Effective Diffusion Transformer Architecture for Image Super-Resolution (AAAI 2025 · similarity 0.88)
- Transformer-Based Selective Super-resolution for Efficient Image Refinement (AAAI 2024 · similarity 0.87)
- CFAT: Unleashing Triangular Windows for Image Super-resolution (CVPR 2024 · similarity 0.87)
- Transcending the Limit of Local Window: Advanced Super-Resolution Transformer with Adaptive Token Dictionary (CVPR 2024 · 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.
