Research map / Video / Video-language understanding
VideoBrain: Learning Adaptive Frame Sampling for Long Video Understanding
Junbo Zou, Ziheng Huang, Shengjie Zhang, Liwen Zhang, Weining 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 CLIP agent (ablation) | VideoBrain · LVBench | accuracy: 36.3 -> 41.3 | Improved |
| add Uniform agent (ablation) | VideoBrain · LVBench | accuracy: 39.8 -> 41.3 | Improved |
| add RL with behavior-aware rewards (ablation) | VideoBrain · LVBench | accuracy: 39.1 -> 41.3 | Improved |
| add SFT cold start (ablation) | VideoBrain · LVBench | accuracy: 35.5 -> 41.3 | Improved |
| classify SFT data into Direct/Adaptive/Active (ablation) | VideoBrain · LVBench | accuracy: 36.2 -> 41.3 | Improved |
| classify RL data (ablation) | VideoBrain · LVBench | accuracy: 40.8 -> 41.3 | Improved |
| add behavior-aware reward shaping (ablation) | VideoBrain · LVBench | accuracy: 39.0 -> 41.3 | Improved |
| add format check reward (ablation) | VideoBrain · LVBench | accuracy: 39.7 -> 41.3 | Improved |
| adaptive frame sampling agents on Qwen3-VL-8B | Qwen3-VL-8B-Instruct · LongVideoBench | accuracy: 45.2 -> 53.3 | Improved |
| apply VideoBrain to short videos without fine-tuning | Qwen3-VL-8B-Instruct · DREAM-1K | precision: 33.6 -> 38.7 | Improved |
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.
- FrameThinker — FrameThinker: Learning to Think with Long Videos via Multi-Turn Frame Spotlighting (ICLR 2026) · better on 3 of 3 shared results
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
- BOLT: Boost Large Vision-Language Model Without Training for Long-form Video Understanding (CVPR 2025 · similarity 0.90)
- MotionBench: Benchmarking and Improving Fine-grained Video Motion Understanding for Vision Language Models (CVPR 2025 · similarity 0.89)
- Adaptive Keyframe Sampling for Long Video Understanding (CVPR 2025 · similarity 0.89)
- ReAgent-V: A Reward-Driven Multi-Agent Framework for Video Understanding (NeurIPS 2025 · similarity 0.89)
- ReWind: Understanding Long Videos with Instructed Learnable Memory (CVPR 2025 · similarity 0.89)
- PerceptionLM: Open-Access Data and Models for Detailed Visual Understanding (NeurIPS 2025 · similarity 0.89)
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.
