Research map / Video / Video-language understanding
Asynchronous Temporal Modeling with Two-Agent Framework for Streaming Dense Video Captioning
Yolo Y. Tang, Chao Huang, Susan Liang, Jing Bi, Yicheng Wang, Daiki Shimada, Chenliang Xu
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 Oracle agent for boundary detection to fixed decoding points (ablation) | Takusen · ActivityNet Captions | CIDEr: 30.4 -> 43.7 | Improved |
| add fixed decoding points alongside Oracle boundaries (ablation) | Takusen (CM2 Oracle) · ActivityNet Captions | CIDEr: 12.3 -> 43.7 | Improved |
| add Listener LMM on top of Oracle model alone (ablation) | Takusen (CM2) · ActivityNet Captions | CIDEr: 33.1 -> 43.7 | Improved |
| use CM2 instead of Videollama3-2b as Oracle without FDP (ablation) | Takusen · ActivityNet Captions | CIDEr: 20.3 -> 12.3 | Hurt |
| provide ground-truth event boundaries instead of Oracle predictions (ablation) | Takusen · ActivityNet Captions | CIDEr: 43.7 -> 107.5 | Improved |
| increase fixed decoding points from 2 to 16 without Oracle (ablation) | Listener only · YouCook2 | recall: 5.3 -> 35.9 | Improved |
| increase fixed decoding points from 2 to 16 without Oracle (ablation) | Listener only · YouCook2 | precision: 21.5 -> 18.9 | Hurt |
| increase fixed decoding points beyond 10 (ablation) | Listener only · YouCook2 | F1: CIDEr 17.9 -> 13.2 at 16 FDPs; F1 24.5 -> 24.8 at 16 FDPs | Mixed |
| use Takusen in streaming setting instead of Streaming Vid2Seq | streaming dense video captioning · YouCook2 | CIDEr: 32.9 -> 40.7 | Improved |
| use Takusen instead of Streaming Vid2Seq on METEOR | streaming dense video captioning · ActivityNet Captions | METEOR: 10.0 -> 9.7 | Hurt |
Closest papers by meaning
- An Efficient Framework for Dense Video Captioning (AAAI 2020 · similarity 0.89)
- Parallelized Autoregressive Decoding for Omni-Modal Dense Video Captioning (ECCV 2026 · similarity 0.89)
- Streaming Dense Video Captioning (CVPR 2024 · similarity 0.88)
- TimeChat: A Time-sensitive Multimodal Large Language Model for Long Video Understanding (CVPR 2024 · similarity 0.87)
- SVBench: A Benchmark with Temporal Multi-Turn Dialogues for Streaming Video Understanding (ICLR 2025 · similarity 0.86)
- CapsFusion: Rethinking Image-Text Data at Scale (CVPR 2024 · 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.
