Research map / Video / Video segmentation and tracking
Neural Video Compression with Reference Hierarchy
Chuanbo Tang, Zhuoyuan Li, Li Li, Dong Liu, Feng Wu
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 feature-wise inter-frame prediction (low-rank reference feature for motion estimation) (ablation) | DCVC-DC baseline · HEVC class datasets, IP32 | BD-rate (%): 0.0 -> -3.9 | Improved |
| add hierarchical reference structure (HRS) on top of FIP (ablation) | DCVC-DC + FIP · HEVC class datasets, IP32 | BD-rate (%): -3.9 -> -7.4 | Improved |
| add reference balance loss (ablation) | DCVC-DC + FIP + HRS · HEVC class datasets, IP32 | BD-rate (%): -7.4 -> -8.8 | Improved |
| add long-sequence (32-frame) training to baseline (ablation) | DCVC-DC baseline · HEVC class datasets, IP32 | BD-rate (%): 0.0 -> -4.3 | Improved |
| add long-sequence training to reference hierarchy (ablation) | DRHVC · HEVC class datasets, IP32 | BD-rate (%): -8.8 -> -13.4 | Improved |
| use DRHVC instead of DCMVC | DCMVC · 7 datasets, intra-period -1 | BD-rate vs VTM (%): -20.4 -> -26.0 | Improved |
| use DRHVC instead of DCVC-DC | DCVC-DC · 7 datasets, intra-period -1 | BD-rate vs VTM (%): 2.2 -> -26.0 | Improved |
| use DRHVC instead of DCMVC on USTC-TD | DCMVC · USTC-TD, intra-period -1 | BD-rate vs VTM (%): -6.3 -> -5.0 | Hurt |
| use DRHVC instead of DCMVC on USTC-TD (IP32) | DCMVC · USTC-TD, intra-period 32 | BD-rate vs VTM (%): 0.9 -> 1.7 | Hurt |
| add reference hierarchy to DCVC-DC | DCVC-DC · 1080p complexity | encoding time (ms): 663 -> 697 | Hurt |
| add reference hierarchy to DCVC-DC | DCVC-DC · 1080p complexity | decoding time (ms): 557 -> 558 | No clear effect |
Closest papers by meaning
- Neural Video Compression with Context Modulation (CVPR 2025 · similarity 0.91)
- Neural B-frame Video Compression with Bi-directional Reference Harmonization (NeurIPS 2025 · similarity 0.89)
- Towards Practical Real-Time Neural Video Compression (CVPR 2025 · similarity 0.88)
- Neural Video Compression with Feature Modulation (CVPR 2024 · similarity 0.87)
- NVRC: Neural Video Representation Compression (NeurIPS 2024 · similarity 0.86)
- PNVC: Towards Practical INR-based Video Compression (AAAI 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.
