Research map / Image generation / Super-resolution
Bridging the Perception Gap in Image Super-Resolution Evaluation
Shaolin Su, Josep M. Rocafort, Danna Xue, David Serrano-Lozano, Lei Sun, Javier Vazquez-Corral
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 |
|---|---|---|---|
| train IQA models with RQI relative-discrepancy scheme instead of traditional FR training (ablation) | AHIQ/MANIQA/TOPIQ · SR user-study datasets | SRCC: MANIQA on PIPAL, DIV2K: 0.624 -> 0.744 | Improved |
| train TOPIQ on PieAPP with RQI (ablation) | TOPIQ · Set5&Set14 user study | SRCC: 0.035 -> 0.010 | Hurt |
| use RQI instead of distortion-oriented FR metrics (SSIM, PSNR) | SR evaluation · DIV2K user study | SRCC: -0.348 -> 0.744 (SSIM vs RQI) | Improved |
| add RQI as auxiliary loss to SeeSR | SeeSR · LSDIR val | Clip-IQA: 0.6364 -> 0.6655 | Improved |
| add RQI as auxiliary loss to PiSA-SR | PiSA-SR · LSDIR val | LPIPS: 0.2382 -> 0.2320 | Improved |
| add RQI as auxiliary loss to SwinIR | SwinIR · LSDIR val | DeQA: 4.0156 -> 4.0762 | Improved |
| add RQI loss to PiSA-SR | PiSA-SR · DIV2K-wild val | PSNR: 23.8365 -> 23.5346 | Hurt |
| use RQI instead of AFINE as auxiliary SR loss | SeeSR · LSDIR val | SSIM: 0.5774 -> 0.6041 | Improved |
| use RQI vs DeQA-Score on small SR benchmark | SR evaluation · Set5&Set14 user study | SRCC: 0.699 -> 0.664 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SeeSR — SeeSR: Towards Semantics-Aware Real-World Image Super-Resolution (CVPR 2024) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Augmenting Perceptual Super-Resolution via Image Quality Predictors (CVPR 2025 · similarity 0.93)
- Evaluation of Visual Processing Should Be Human-Centered, Not Metric-Centered (NeurIPS 2026 · similarity 0.89)
- DP²O-SR: Direct Perceptual Preference Optimization for Real-World Image Super-Resolution (NeurIPS 2025 · similarity 0.88)
- SEAL: A Framework for Systematic Evaluation of Real-World Super-Resolution (ICLR 2024 · similarity 0.88)
- Toward Generalized Image Quality Assessment: Relaxing the Perfect Reference Quality Assumption (CVPR 2025 · similarity 0.88)
- Pixel-level and Semantic-level Adjustable Super-resolution: A Dual-LoRA Approach (CVPR 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.
