Research map / Image generation / Super-resolution
Reference-based Image and Video Super-Resolution via $C^{2}$-Matching
Yuming Jiang, Kelvin C. K. Chan, Xintao Wang, Chen Change Loy, Ziwei Liu
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 dynamic aggregation module to base model (ablation) | C2-Matching Ref image SR · CUFED5 | PSNR: +0.08dB | Improved |
| replace fixed VGG feature matching with contrastive correspondence network (ablation) | C2-Matching Ref image SR · CUFED5 | PSNR: about 0.2dB | Improved |
| add teacher-student correlation distillation (ablation) | C2-Matching Ref image SR · CUFED5 full | PSNR: +0.09dB | Improved |
| add teacher-student correlation distillation (ablation) | C2-Matching Ref image SR · CUFED5 texture subset | PSNR: +0.16dB | Improved |
| use C2-Matching instead of optical flow for reference alignment (ablation) | C2-Matching Ref VSR · REDS4 | PSNR: 31.85 -> 32.05 | Improved |
| add attention fusion of forward, backward and reference features (ablation) | C2-Matching Ref VSR · REDS4 | PSNR: 31.96 -> 32.05 | Improved |
| give a very similar rather than irrelevant reference image (ablation) | C2-Matching Ref VSR · REDS4 last 50 frames | PSNR: 31.38 -> 32.05 | Improved |
| give an irrelevant reference image versus none (ablation) | C2-Matching Ref VSR · REDS4 last 50 frames | PSNR: 31.40 -> 31.38 | No clear effect |
| use light version (fewer residual blocks, drop first dynamic aggregation) (ablation) | C2-Matching · CUFED5 | PSNR: 28.24 -> 28.14 | Hurt |
| train with GAN loss instead of reconstruction loss only | C2-Matching · CUFED5 | PSNR: 28.24 -> 27.16 | Hurt |
| use C2-Matching instead of optical flow for alignment | C2-Matching Ref VSR · inference time | seconds per video: 82.44 -> 163.11 | Hurt |
Closest papers by meaning
- Robust Reference-Based Super-Resolution via C2-Matching (CVPR 2021 · similarity 0.98)
- Image Super-Resolution by Neural Texture Transfer (CVPR 2019 · similarity 0.91)
- Task Decoupled Framework for Reference-Based Super-Resolution (CVPR 2022 · similarity 0.91)
- LMR: A Large-Scale Multi-Reference Dataset for Reference-Based Super-Resolution (ICCV 2023 · similarity 0.90)
- Sequence Matters: Harnessing Video Models in 3D Super-Resolution (AAAI 2025 · similarity 0.87)
- Building Bridges across Spatial and Temporal Resolutions: Reference-Based Super-Resolution via Change Priors and Conditional Diffusion Model (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.
