Research map / Video / Video segmentation and tracking
Conditional Entropy Coding for Efficient Video Compression
Jerry Liu, Shenlong Wang, Wei-Chiu Ma, Meet Shah, Rui Hu, Pranaab Dhawan, Raquel Urtasun
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 internal learning of latent codes at inference (ablation) | conditional entropy video compressor · UVG rate-distortion | bitrate: ~10% additional bitrate savings | Improved |
| add the conditional entropy model on top of single-image compression (ablation) | single-image Liu model · NorthAmerica | bitrate: bitrate improvements of 20-50% | Improved |
| add internal learning on NorthAmerica (ablation) | conditional entropy video compressor · NorthAmerica | bitrate: additional ~10% improvement | Improved |
| add conditional entropy model at lower framerates (ablation) | single-image model · UVG and CDVL, framerates 1/3, 1/6, 1/10 | bitrate: 20%-50% reduction of bitrate | Improved |
| add internal learning at lower framerates (ablation) | conditional entropy video compressor · UVG and CDVL, framerates 1/3, 1/6, 1/10 | bitrate: additional 10-20% reduction in bitrate | Improved |
| increase internal learning gradient steps from 1 to 100 (ablation) | conditional entropy video compressor · UVG ShakeNDry (first 50 frames) | rate-distortion: saturates after 30 gradient steps | Mixed |
| increase H.265/H.264 GoP size from 12 to full video length (ablation) | H.265 / H.264 · UVG | rate-distortion: marginal gap between GoP 12 and default; thin gap to max GoP | No clear effect |
| use single-image compression instead of H.265 veryslow at lower framerates | single-image compressor · UVG lower framerates | rate-distortion: matches H.265 veryslow | No clear effect |
| independent frame encoder lacks high-frequency capture on VTL | conditional entropy video compressor · VTL | rate-distortion vs video codecs | Hurt |
Closest papers by meaning
- Efficient Learned Image Compression without Entropy Coding (ICML 2026 · similarity 0.89)
- Video Compression With Entropy-Constrained Neural Representations (CVPR 2023 · similarity 0.89)
- Ultra-Fast Neural Video Compression (CVPR 2026 · similarity 0.89)
- Learned Image Compression With Discretized Gaussian Mixture Likelihoods and Attention Modules (CVPR 2020 · similarity 0.89)
- Deep Contextual Video Compression (NeurIPS 2021 · similarity 0.89)
- Neural Video Compression with Feature Modulation (CVPR 2024 · similarity 0.88)
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.
