arXiv cs.AI· Runquan Gui, Hanzhu Chen, Zehao Wang, Hanxin Zhu, Xin Li, Zhibo Chen·· 9 天前AI 评分30
VAMR：面向高效长视频理解的多问题智能体推理
VAMR: Multi-Question Agentic Reasoning for Efficient Long-Form Video Understanding
AI 导读
VAMR 通过一条共享的工具调用轨迹统一处理同一长视频的多个问题，避免每个问题各自重启视频探索与记忆构建。在 LVBench、Video-Holmes 和 LongVideoBench 上，VAMR 总体准确率最高且迭代方法中推理轮数最少；LVBench 上达 62.1%，超 VideoARM 4.3 分，推理轮数与处理帧数分别减少 85.9% 和 61.4%。
正文 · 原文
Abstract:Long-form video understanding often involves multiple questions about different aspects of the same recording. Yet existing video agents typically process each question through an isolated tool-use trajectory. This repeatedly restarts video exploration and memory construction, missing opportunities to acquire evidence jointly and progressively build a shared understanding that supports the complete question set. We introduce \textbf{VAMR} (\textbf{V}ideo \textbf{A}gent for \textbf{M}ulti-Question \textbf{R}easoning), which coordinates all questions about a video through one shared tool-use trajectory. At each round, a persistent policy model can invoke tools for one or more unresolved questions and submit answers for questions with sufficient evidence. Question-conditioned visual perception retrieves fine-grained clues for several questions in one call, while layered multi-question memory integrates reusable context into a shared video story and preserves separate evidence for individual questions. After supervised fine-tuning initializes this interaction protocol, we propose question-horizon policy optimization (\qhpo) to optimize shared trajectories in which questions progress and finish at different rounds. Specifically, a question-level critic estimates the value of each active question, while round alignment maps each question advantage to the rounds that directly serve it before the aligned advantages are aggregated to optimize the shared actor. Across LVBench, Video-Holmes, and LongVideoBench, VAMR achieves the highest accuracy overall and the fewest reasoning rounds among iterative methods. On LVBench, it reaches 62.1\% accuracy, exceeding VideoARM by \textbf{4.3} points while reducing reasoning rounds and processed frames by \textbf{85.9\%} and \textbf{61.4\%}.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.11171 [cs.CV] |
| (or arXiv:2610.11171v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.11171 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Runquan Gui [view email]
[v1]
Thu, 8 Oct 2026 03:26:54 UTC (275 KB)
来源：arXiv cs.AI · arxiv.org