Research map / Video / Video segmentation and tracking
Compression-Aware Video Super-Resolution
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Condition VSR features on an estimated compression representation (ablation) | BasicVSR · Vid4 CRF15, CRF25, and CRF35 | PSNR (dB): 26.76 to 27.25; 24.54 to 25.41; 22.06 to 22.74 | Helped |
| Use metadata-assisted alignment (ablation) | CAM-enhanced BasicVSR · Vid4 CRF15, CRF25, and CRF35 | PSNR (dB): 27.25 to 27.40; 25.41 to 25.60; 22.74 to 22.80 | Helped |
| Use metadata-assisted hidden-state updating (ablation) | CAM + meta-assisted alignment · Vid4 CRF15, CRF25, and CRF35 | PSNR (dB): 27.40 to 27.42; 25.60 to 25.65; 22.80 to 22.84 | Helped |
| Train the compression encoder with ranking loss rather than contrastive loss (ablation) | Compression-aware modulation · Vid4 CRF15, CRF25, and CRF35 | PSNR (dB): 27.20 to 27.32; 25.49 to 25.58; 22.62 to 22.77 | Helped |
| Add frame-type-varied pairs to compression-encoder training (ablation) | Compression-aware modulation with ranking loss · Vid4 CRF15, CRF25, and CRF35 | PSNR (dB): 27.32 to 27.42; 25.58 to 25.65; 22.77 to 22.84 | Helped |
| Use optical flow for alignment instead of motion vectors (ablation) | Meta-assisted alignment · Vid4 CRF15 and CRF25 versus CRF35 | PSNR (dB): 27.36 to 27.53; 25.57 to 25.69; 22.79 to 22.74 | Mixed |
| Refine codec motion vectors using frames and residual maps (ablation) | Meta-assisted alignment · Vid4 CRF15, CRF25, and CRF35 | PSNR (dB): 27.36 to 27.42; 25.57 to 25.65; 22.79 to 22.84 | Helped |
| Use compression-aware conditioning and compressed-video metadata | CAVSR · Vid4 CRF15 and CRF25 | PSNR (dB): CRF15: 27.42 vs 27.50; CRF25: 25.65 vs 25.51 | Mixed |
| Use compression-aware conditioning and compressed-video metadata | CAVSR · Vid4 CRF35 | PSNR (dB): 22.79 to 22.84 | Helped |
Closest papers by meaning
- Compressed-Domain-Aware Online Video Super-Resolution (CVPR 2026 · similarity 0.88)
- Arbitrary-Scale Video Super-resolution Guided by Dynamic Context (AAAI 2024 · similarity 0.85)
- NegVSR: Augmenting Negatives for Generalized Noise Modeling in Real-world Video Super-Resolution (AAAI 2024 · similarity 0.85)
- VideoGigaGAN: Towards Detail-rich Video Super-Resolution (CVPR 2025 · similarity 0.85)
- DOVE: Efficient One-Step Diffusion Model for Real-World Video Super-Resolution (NeurIPS 2025 · similarity 0.84)
- PatchVSR: Breaking Video Diffusion Resolution Limits with Patch-wise Video Super-Resolution (CVPR 2025 · similarity 0.84)
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.