StreamArena：面向连续交互与长时程智能体流媒体视频理解的基准与架构
原标题：StreamArena: Toward Continuous, Interactive, and Long-Horizon Agentic Streaming Video Understanding
AI 摘要
StreamArena是一个用于小时级、交互式流媒体视频理解的基准测试，包含243个平均时长88.8分钟的视频和3646个开放式问答对，评估实时感知、历史回顾、主动交互和多模态工具使用能力。研究发现现有方法在连续交互和长时程多模态理解之间存在矛盾，为此提出StreamMind架构，通过前端工作者处理延迟关键任务、后端工作者构建持久多模态记忆，在四项能力上均优于现有基线并降低查询响应延迟。
正文节选
StreamArena: Toward Continuous, Interactive, and Long-Horizon Agentic Streaming Video Understanding Abstract Deploying autonomous multimodal agents in continuous, real-world environments requires them to ingest unbounded audio-visual streams and maintain hour-scale memory. However, current evaluations predominantly rely on brief clips and multiple-choice formats. This design allows minimal baselines that process only the last four frames to match or surpass complex streaming models, while answer