Research map / Video / Video-language understanding
CogStream: Context-guided Streaming Video Question Answering
Zicheng Zhao, Kangyu Wang, Shijie Li, Rui Qian, Weiyao Lin, Huabin 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 |
|---|---|---|---|
| add temporal-semantic clustering (ablation) | CogReasoner (fine-tuned) · CogStream | avg score: 71.48 -> 72.26 | Improved |
| summarize low-relevance events instead of discarding them (ablation) | CogReasoner (fine-tuned) · CogStream | avg score: 70.52 -> 72.26 | Improved |
| add full visual stream compression vs baseline (ablation) | CogReasoner (fine-tuned) · CogStream | avg score: 70.64 -> 72.26 | Improved |
| use all historical QA instead of none (ablation) | CogReasoner (fine-tuned) · CogStream | avg score: 66.04 -> 70.48 | Improved |
| retrieve relevant historical QA instead of using all context (ablation) | CogReasoner (fine-tuned) · CogStream | avg score: 70.48 -> 72.26 | Improved |
| use retrieved context instead of all context on Basic QA (ablation) | CogReasoner (fine-tuned) · CogStream Basic | score: 75.20 -> 75.30 | No clear effect |
| use all history vs none on Basic QA (ablation) | CogReasoner (fine-tuned) · CogStream Basic | score: 75.60 -> 75.20 | No clear effect |
| add trained HDR module (ablation) | VideoLLaMA2 · CogStream | avg score: 50.72 -> 55.16 | Improved |
| add trained HDR module (ablation) | MiniCPM-V-2.6 · CogStream | avg score: 66.84 -> 67.78 | Improved |
| let the model's own LLM do historical retrieval (ablation) | VideoLLaMA2 · CogStream | avg score: 50.72 -> 50.52 | Hurt |
| let the model's own LLM do historical retrieval (ablation) | MiniCPM-V-2.6 · CogStream | avg score: 66.84 -> 65.46 | Hurt |
| add 30% irrelevant QA noise with all-context (ablation) | CogReasoner (fine-tuned) · CogStream | avg score: 70.48 -> 69.74 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Flash-VStream — Flash-VStream: Efficient Real-Time Understanding for Long Video Streams (ICCV 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Streaming Video Understanding and Multi-round Interaction with Memory-enhanced Knowledge (ICLR 2025 · similarity 0.89)
- SVBench: A Benchmark with Temporal Multi-Turn Dialogues for Streaming Video Understanding (ICLR 2025 · similarity 0.89)
- Streaming Long Video Understanding with Large Language Models (NeurIPS 2024 · similarity 0.89)
- LongVideoBench: A Benchmark for Long-context Interleaved Video-Language Understanding (NeurIPS 2024 · similarity 0.88)
- Temporal Chain of Thought: Long-Video Understanding by Thinking in Frames (NeurIPS 2025 · similarity 0.88)
- VideoLLM-online: Online Video Large Language Model for Streaming Video (CVPR 2024 · similarity 0.87)
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.
