Research map / Video / Video segmentation and tracking
Real-Time Neural Video Compression with Unified Intra and Inter Coding
Hui Xiang, Yifan Bian, Li Li, Jingran Wu, Xianguo Zhang, Dong 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 |
|---|---|---|---|
| unify intra and inter coding in one model (no refresh) (ablation) | DCVC-RT style codec · HEVC B-E average | BD-rate (%): 93.9 -> 29.0 | Improved |
| add simultaneous two-frame compression (ablation) | unified model, no refresh · HEVC B-E average | BD-rate (%): 29.0 -> 5.3 | Improved |
| add hybrid reference training (ablation) | unified two-frame model, no refresh · HEVC B-E average | BD-rate (%): 5.3 -> 0.0 | Improved |
| add periodic refresh (64) to divided I/P model (ablation) | divided model · HEVC B-E average | BD-rate (%): 93.9 -> 33.8 | Improved |
| add periodic refresh (64) to unified model (ablation) | unified model · HEVC B-E average | BD-rate (%): 29.0 -> 66.4 | Hurt |
| add refresh (64) to full model (ablation) | UI2C full · HEVC B-E average | BD-rate (%): 0.0 -> 33.7 | Hurt |
| replace DCVC-RT with UI2C | DCVC-RT · average over test sets | BD-rate (%): 0.0 -> -12.1 | Improved |
| replace DCVC-RT with UI2C on short sequences | DCVC-RT · MCL-JCV | BD-rate (%): 0.0 -> 1.1 | Hurt |
| two-frame joint coding raises encoding speed and cuts params with single model | UI2C vs DCVC-RT · 1080p RTX 3090 | encoding fps: 56.8 -> 65.1 | Improved |
| unified model inserts blank-reference intra coding vs DCVC-RT P-model intra | first two frames · HEVC B, E, UVG first two frames | RD performance | Improved |
Closest papers by meaning
- Towards Practical Real-Time Neural Video Compression (CVPR 2025 · similarity 0.94)
- NVRC: Neural Video Representation Compression (NeurIPS 2024 · similarity 0.91)
- Neural Video Compression with Context Modulation (CVPR 2025 · similarity 0.90)
- PNVC: Towards Practical INR-based Video Compression (AAAI 2025 · similarity 0.90)
- Neural Video Compression with Feature Modulation (CVPR 2024 · similarity 0.90)
- Combining Frame and GOP Embeddings for Neural Video Representation (CVPR 2024 · similarity 0.89)
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.
