Research map / Image generation / Super-resolution
Dynamic High-Pass Filtering and Multi-Spectral Attention for Image Super-Resolution
Salma Abdel Magid, Yulun Zhang, Donglai Wei, Won-Dong Jang, Zudi Lin, Yun Fu, Hanspeter Pfister
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 |
|---|---|---|---|
| use 8 frequency components instead of 1 in channel attention (ablation) | MMCA · Urban100 x4 | PSNR (dB): 26.24 -> 26.39 | Improved |
| use 16 instead of 8 frequency components (ablation) | MMCA · Urban100 x4 | PSNR (dB): 26.39 -> 26.33 | No clear effect |
| use 2 frequency components instead of 1 (ablation) | MMCA · Urban100 x4 | PSNR (dB): 26.24 -> 26.36 | Improved |
| add one HPF layer to the residual block (ablation) | RCAN residual block · Manga109 x4 | PSNR (dB): 30.65 -> 30.82 | Improved |
| add a second HPF layer after the second convolution (ablation) | RCAN residual block · Manga109 x4 | PSNR (dB): 30.82 -> 30.74 | Hurt |
| use 16 HPF groups instead of 2 (ablation) | HPF module · Manga109 x4 | PSNR (dB): 30.82 -> 30.88 | Improved |
| use 4 HPF groups instead of 2 (ablation) | HPF module · Manga109 x4 | PSNR (dB): 30.82 -> 30.79 | No clear effect |
| use 8 HPF groups instead of 2 (ablation) | HPF module · Manga109 x4 | PSNR (dB): 30.82 -> 30.82 | No clear effect |
| replace SENet with MMCA channel attention (ablation) | RCAN · Urban100 x4 | PSNR (dB): 26.24 -> 26.39 | Improved |
| replace FcaNet with MMCA channel attention (ablation) | RCAN · Urban100 x4 | PSNR (dB): 26.29 -> 26.39 | Improved |
| replace SENet with MMCA channel attention (ablation) | RCAN · Manga109 x4 | PSNR (dB): 30.65 -> 30.77 | Improved |
| replace FcaNet with MMCA channel attention (ablation) | RCAN · Manga109 x4 | PSNR (dB): 30.67 -> 30.77 | Improved |
Closest papers by meaning
- Single Image Super-Resolution via a Holistic Attention Network (ECCV 2020 · similarity 0.91)
- Learning Frequency-Aware Dynamic Network for Efficient Super-Resolution (ICCV 2021 · similarity 0.90)
- Image Super-Resolution Using Very Deep Residual Channel Attention Networks (ECCV 2018 · similarity 0.90)
- Exploring the Low-Pass Filtering Behavior in Image Super-Resolution (ICML 2024 · similarity 0.89)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.87)
- ResDiff: Combining CNN and Diffusion Model for Image Super-resolution (AAAI 2024 · similarity 0.87)
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.
