Research map / Video / Video segmentation and tracking
COMISR: Compression-Informed Video Super-Resolution
Yinxiao Li, Pengchong Jin, Feng Yang, Ce Liu, Ming-Hsuan Yang, Peyman Milanfar
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 bi-directional recurrent module (ablation) | Recur baseline · Vid4 CRF25 (aug CRF15-25) | PSNR (Y-channel): 24.23 -> 24.74 | Improved |
| add detail-aware flow estimation (ablation) | Recur + a · Vid4 CRF25 (aug CRF15-25) | PSNR (Y-channel): 24.74 -> 24.92 | Improved |
| add Laplacian enhancement module (ablation) | Recur + ab · Vid4 CRF25 (aug CRF15-25) | PSNR (Y-channel): 24.92 -> 24.97 | Improved |
| add bi-directional recurrent module (ablation) | Recur baseline · Vid4 CRF25 (no compression aug) | PSNR (Y-channel): 23.97 -> 24.24 | Improved |
| add compressed images (CRF15-25) in last 20% of training (ablation) | COMISR · Vid4 CRF25 | PSNR (Y-channel): 24.31 -> 24.97 | Improved |
| add compression augmentation in training (ablation) | COMISR · Vid4 uncompressed | PSNR (Y-channel): 27.48 -> 27.31 | Hurt |
| apply STDF video denoiser before VSR | EDVR · Vid4 CRF25 | PSNR (Y-channel): 24.45 -> 22.56 | Hurt |
| apply STDF video denoiser before VSR | TecoGan · Vid4 CRF25 | PSNR (Y-channel): 23.94 -> 22.25 | Hurt |
| apply STDF video denoiser before VSR | RSDN · Vid4 CRF25 | PSNR (Y-channel): 24.06 -> 22.19 | Hurt |
| compare COMISR with prior VSR on compressed REDS4 | COMISR vs EDVR · REDS4 CRF25 (RGB) | PSNR (Y-channel): 25.98 -> 26.47 | Improved |
| compare COMISR with RSDN on uncompressed Vid4 | COMISR vs RSDN · Vid4 uncompressed | PSNR (Y-channel): 27.92 -> 27.31 | Hurt |
Later papers that compared with this method
2 later accepted papers report COMISR as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- Learning Degradation-Robust Spatiotemporal Frequency-Transformer for Video Super-Resolution (TPAMI 2023 · better on 6 of 6 shared results)
- Learning Spatiotemporal Frequency-Transformer for Compressed Video Super-Resolution (ECCV 2022 · better on 5 of 7 shared results)
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.
- MuCAN — MuCAN: Multi-Correspondence Aggregation Network for Video Super-Resolution (ECCV 2020) · better on 4 of 4 shared results
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
- Compression-Aware Video Super-Resolution (CVPR 2023 · similarity 0.92)
- Exploring Real-Time Super-Resolution: Benchmarking and Fine-Tuning for Streaming Content (ICLR 2026 · similarity 0.89)
- Compressed-Domain-Aware Online Video Super-Resolution (CVPR 2026 · similarity 0.89)
- Learning Data-Driven Vector-Quantized Degradation Model for Animation Video Super-Resolution (ICCV 2023 · similarity 0.88)
- SparkVSR: Interactive Video Super-Resolution via Sparse Keyframe Propagation (ECCV 2026 · similarity 0.88)
- SkipSR: Faster Super-Resolution with Token Skipping (NeurIPS 2026 · 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.
