Research map / Image generation / Super-resolution
Reference-Based Image Super-Resolution with Deformable Attention Transformer
Jiezhang Cao, Jingyun Liang, Kai Zhang, Yawei Li, Yulun Zhang, Wenguan Wang, Luc Van Gool
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 reference-based deformable attention (RDA) instead of feature warping on the most relevant correspondence (ablation) | DATSR · CUFED5 | PSNR/SSIM | Improved |
| add residual feature aggregation (RFA) instead of plain CNNs (ablation) | DATSR · CUFED5 | PSNR/SSIM | Improved |
| replace RFA with CNNs but keep RDA vs C2-Matching (ablation) | DATSR with CNNs · CUFED5 | PSNR/SSIM | Improved |
| train DATSR with adversarial loss instead of reconstruction loss only | DATSR · CUFED5 | PSNR: 28.72 -> 27.95 | Hurt |
| DATSR vs C2-Matching, reconstruction loss | reference-based SR · CUFED5 | PSNR: 28.24 -> 28.72 | Improved |
| DATSR vs C2-Matching on L4 least similar references | reference-based SR · CUFED5 L4 | PSNR: 26.85 -> 26.96 | Improved |
| DATSR vs C2-Matching on L1 most similar references | reference-based SR · CUFED5 L1 | PSNR: 28.11 -> 28.50 | Improved |
| DATSR vs C2-Matching, all losses | reference-based SR · Manga109 vs WR-SR | PSNR: 27.80 -> 27.87 | Improved |
| DATSR (random reference) vs SwinIR on Urban100 | reference-based SR · Urban100 | PSNR: 26.26 -> 26.52 | Improved |
| single reference DATSR vs multi-reference CIMR-SR and AMRSR | DATSR · CUFED5 | PSNR | Improved |
Closest papers by meaning
- Image Super-Resolution by Neural Texture Transfer (CVPR 2019 · similarity 0.92)
- Task Decoupled Framework for Reference-Based Super-Resolution (CVPR 2022 · similarity 0.92)
- LMR: A Large-Scale Multi-Reference Dataset for Reference-Based Super-Resolution (ICCV 2023 · similarity 0.91)
- MASA-SR: Matching Acceleration and Spatial Adaptation for Reference-Based Image Super-Resolution (CVPR 2021 · similarity 0.90)
- Robust Reference-Based Super-Resolution via C2-Matching (CVPR 2021 · similarity 0.90)
- 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.
