Research map / Image generation / Super-resolution
Context Reasoning Attention Network for Image Super-Resolution
Yulun Zhang, Donglai Wei, Can Qin, Huan Wang, Hanspeter Pfister, Yun Fu
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 |
|---|---|---|---|
| replace residual blocks with RCAB (ablation) | EDSR baseline · DIV2K val x2 EDSR baseline | PSNR: 37.15 -> 37.19 | Improved |
| replace residual blocks with CRAB without CDRR (ablation) | EDSR baseline · DIV2K val x2 EDSR baseline | PSNR: 37.15 -> 37.28 | Improved |
| add context descriptor relationship reasoning (CDRR) to CRAB (ablation) | CRAB · DIV2K val x2 EDSR baseline | PSNR: 37.28 -> 37.34 | Improved |
| combine spatial and channel interaction components (all) (ablation) | CRAB · DIV2K val x2 EDSR baseline | PSNR: 37.22 -> 37.34 | Improved |
| insert one CRAB at 1-st RB (ablation) | EDSR baseline · DIV2K val x2 EDSR baseline | PSNR: 37.15 -> 37.17 | Improved |
| insert one CRAB at 4-th RB (ablation) | EDSR baseline · DIV2K val x2 EDSR baseline | PSNR: 37.15 -> 37.18 | Improved |
| insert one CRAB at 8-th RB (ablation) | EDSR baseline · DIV2K val x2 EDSR baseline | PSNR: 37.15 -> 37.22 | Improved |
| insert one CRAB at 16-th RB (highest level) (ablation) | EDSR baseline · DIV2K val x2 EDSR baseline | PSNR: 37.15 -> 37.27 | Improved |
| add self-ensemble (CRAN+) (ablation) | CRAN · Urban100 x4 (BI) | PSNR: 27.13 -> 27.30 | Improved |
| use CRAN instead of RCAN | RCAN · Manga109 x4 (BI) | PSNR: 31.22 -> 31.75 | Improved |
| use CRAN instead of CSNLN on Urban100 x4 | CSNLN · Urban100 x4 (BI) | PSNR: 27.22 -> 27.13 | Mixed |
| use CRAN instead of RDN under DN degradation | RDN · Urban100 x3 (DN) | PSNR: 24.92 -> 25.43 | Improved |
Closest papers by meaning
- Second-Order Attention Network for Single Image Super-Resolution (CVPR 2019 · similarity 0.89)
- MR Image Super-Resolution With Squeeze and Excitation Reasoning Attention Network (CVPR 2021 · similarity 0.88)
- Image Super-Resolution Using Very Deep Residual Channel Attention Networks (ECCV 2018 · similarity 0.87)
- Image Super-Resolution Using Deep Convolutional Networks (TPAMI 2016 · similarity 0.87)
- UCAN: Unified Convolutional Attention Network for Expansive Receptive Fields in Lightweight Super-Resolution (CVPR 2026 · similarity 0.87)
- Single Image Super-Resolution via a Holistic Attention Network (ECCV 2020 · 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.
