Research map / Image generation / Super-resolution
Boosting Single Image Super-Resolution via Partial Channel Shifting
Xiaoming Zhang, Tianrui Li, Xiaole Zhao
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 partial channel shifting to DRRN | DRRN · Urban100 x2 | PSNR (dB): 31.32 -> 31.51 | Improved |
| add partial channel shifting to EDSR(S) | EDSR(S) · Urban100 x2 | PSNR (dB): 31.98 -> 32.17 | Improved |
| add partial channel shifting to EDSR(L) | EDSR(L) · Urban100 x2 | PSNR (dB): 32.93 -> 32.99 | Improved |
| add partial channel shifting to EDSR(L) | EDSR(L) · Manga109 x2 | PSNR (dB): 39.10 -> 39.24 | Improved |
| place PCS at front of residual block (position 1) rather than other positions (ablation) | EDSR(S, PCS) · Urban100 x2 | PSNR (dB): 31.99 -> 32.17 | Improved |
| place PCS at position 4 instead of position 1 (ablation) | EDSR(S, PCS) · Urban100 x2 | PSNR (dB): 32.17 -> 32.05 | Hurt |
| vary shifting directions (one to eight directions) (ablation) | DRRN(PCS), EDSR(S, PCS) · B100 x2 | PSNR (dB): 31.99 - 32.02 (DRRN), 32.16 - 32.20 (EDSR(S)) | No clear effect |
| vary permutation of shifted channels (g=1 to 4 groups) (ablation) | DRRN(PCS), EDSR(S, PCS) · B100 x2 | PSNR (dB): 31.98 - 32.01 (DRRN), 32.18 - 32.20 (EDSR(S)) | No clear effect |
| cover eight directions (octa-directional) instead of two (bi-directional) | EDSR(S, PCS) · Urban100 x2 | PSNR (dB) | Hurt |
| increase shifting magnitude to very large values (e.g. 8) | EDSR(S, PCS) · Urban100 x2 | PSNR (dB) | Hurt |
| vary shifted channel proportion gamma | EDSR(L, PCS) · Urban100 x2 | PSNR (dB) | No clear effect |
| use shifting magnitude of 2 | EDSR(S, PCS) · Urban100 x2 | PSNR (dB) | Improved |
Closest papers by meaning
- Spatially-Adaptive Feature Modulation for Efficient Image Super-Resolution (ICCV 2023 · similarity 0.89)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.87)
- FSR: A General Frequency-Oriented Framework to Accelerate Image Super-resolution Networks (AAAI 2023 · similarity 0.87)
- Perception-Oriented Single Image Super-Resolution Using Optimal Objective Estimation (CVPR 2023 · similarity 0.87)
- IAFMNet: Information-Aware Feature Modulation for Efficient Super-Resolution (CVPR 2026 · similarity 0.87)
- SRWarp: Generalized Image Super-Resolution under Arbitrary Transformation (CVPR 2021 · 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.
