Research map / Video / Video-language understanding
Context Guided Transformer Entropy Modeling for Video Compression
Junlong Tong, Wei Zhang, Yaohui Jin, Xiaoyu Shen
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 temporal context resampler (TCR) to CGT (ablation) | CGT · Entropy modeling time | time (ms): 1,305 -> 488 | Improved |
| add TCR to CGT (ablation) | CGT · BD-rate | BD-rate (%): +1.8 | Hurt |
| add TCR to MIMT with random mask (ablation) | MIMT(random) · Entropy modeling time | time (ms): 1,226 -> 427 | Improved |
| add TCR to MIMT with random mask (ablation) | MIMT(random) · BD-rate | BD-rate (%): +2.2 | Hurt |
| add TCR to CGT (ablation) | CGT · Encoding/decoding time | time (ms): 1,682 -> 1,073 encoding; 1,576 -> 984 decoding | Improved |
| use dependency-weighted spatial context assigner instead of min-entropy decoding (ablation) | CGT · MCL-JCV, UVG, HEVC-B average | BD-rate (%): 0 -> -11.2 | Improved |
| use checkerboard decoding instead of min-entropy decoding (ablation) | CGT · average of 3 datasets | BD-rate (%): 0 -> 17.3 | Hurt |
| use block-autoregressive decoding instead of min-entropy decoding (ablation) | CGT · average of 3 datasets | BD-rate (%): 0 -> 19.5 | Hurt |
| set alpha=0 (importance only) instead of alpha=1 (certainty only) (ablation) | CGT · MCL-JCV lambda=256 | PSNR: PSNR 35.3 -> 35.88, Bpp 0.017 -> 0.019 | Mixed |
| use explicit teacher-student modeling instead of random-mask proxy task (ablation) | CGT · MCL-JCV alpha=0.5 | PSNR: 35.58 -> 35.82 | Improved |
| use explicit modeling instead of proxy task (ablation) | CGT · MCL-JCV alpha=0 | Bpp: 0.021 -> 0.019 | Improved |
| replace DMC frame codec with DCVC-DC | CGT · entropy coding | BD-rate | No clear effect |
Closest papers by meaning
- Diversify, Contextualize, and Adapt: Efficient Entropy Modeling for Neural Image Codec (NeurIPS 2024 · similarity 0.86)
- Laplacian-guided Entropy Model in Neural Codec with Blur-dissipated Synthesis (CVPR 2024 · similarity 0.86)
- Compressed Context Memory for Online Language Model Interaction (ICLR 2024 · similarity 0.85)
- Neural Video Compression with Context Modulation (CVPR 2025 · similarity 0.85)
- Causal Context Adjustment Loss for Learned Image Compression (NeurIPS 2024 · similarity 0.85)
- Long Context Compression with Activation Beacon (ICLR 2025 · similarity 0.85)
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.
