Research map / Video / Video segmentation and tracking
RealViformer: Investigating Attention for Real-World Video Super-Resolution
Yuehan Zhang, Angela Yao
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 |
|---|---|---|---|
| use channel attention instead of spatial attention for temporal fusion (ablation) | recurrent real-world VSR · REDS4 with noise and JPEG degradations | PSNR and LPIPS | Improved |
| use channel attention instead of spatial attention for temporal fusion on severely blurred inputs (ablation) | recurrent real-world VSR · REDS4 with blur | PSNR | Hurt |
| use channel attention blocks instead of spatial attention blocks in artifact-free standard VSR (ablation) | standard VSR · REDS (no extra degradation) | SSIM: 0.8432 -> 0.8338 | Hurt |
| use channel attention versus spatial attention (output feature covariance) | feature representation · REDS4 | ac(O): input approximately 0.15 -> 0.87 with channel attention | Hurt |
| add CAF and ICA modules to channel-attention baseline (ablation) | RealViformer · VideoLQ and UDM10 | NRQM and LPIPS: channel correlation 0.436 -> 0.422 | Improved |
| use original channel attention (ch-baseline) in place of spatial-attention baseline (ablation) | RealViformer · VideoLQ and UDM10 | NRQM and LPIPS | Improved |
| add ICA module (squeeze-excite and correlation-based rescaling) (ablation) | RealViformer · model predictions | high-frequency power | Improved |
Closest papers by meaning
- Mitigating Artifacts in Real-World Video Super-resolution Models (AAAI 2023 · similarity 0.90)
- Sequence Matters: Harnessing Video Models in 3D Super-Resolution (AAAI 2025 · similarity 0.87)
- VideoGigaGAN: Towards Detail-rich Video Super-Resolution (CVPR 2025 · similarity 0.87)
- NegVSR: Augmenting Negatives for Generalized Noise Modeling in Real-world Video Super-Resolution (AAAI 2024 · similarity 0.87)
- SAVSR: Arbitrary-Scale Video Super-Resolution via a Learned Scale-Adaptive Network (AAAI 2024 · similarity 0.87)
- One-Step Diffusion for Detail-Rich and Temporally Consistent Video Super-Resolution (NeurIPS 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.
