AI看长视频终于不“失忆”了
现在的AI看几分钟视频还行，但看几小时的电影或监控录像就会“失忆”——要么记不住前面发生了什么，要么被海量画面撑爆算力。这篇论文把“看”和“想”拆开：先让AI像翻书一样一页页看视频，把关键事件和因果关系画成一张“记忆地图”（三层结构：场景→事件→细节），然后让推理模型像侦探一样在地图上搜索线索，而不是从头翻视频。结果：理解长视频的准确率比之前最好的AI高了12.5个百分点，和人类专家的差距缩小到3.7%，而它实际“读”的视频内容只有全片的2%。这不是你明天能用的功能，但它意味着AI终于能看懂《盗梦空间》级别的长叙事——未来可能帮你自动总结会议录像、分析体育比赛、甚至审查监控视频里的异常行为。
📄 原文摘要(英文)
Current Vision-Language Models struggle with hours-long videos because processing full-length visual sequences induces prohibitive token explosion and attention dilution. To overcome this, we introduce MemDreamer to decouple perception and reasoning, shifting long-video understanding into an agentic exploration process. As a plug-and-play framework, it incrementally streams videos to construct a Hierarchical Graph Memory, a top-down three-tier architecture for semantic abstraction, anchored by a foundational graph capturing spatiotemporal and causal relations. During inference, the reasoning model employs agentic tool-augmented retrieval, navigating hierarchies, searching nodes, and traversing logical edges via an Observation-Reason-Action loop. Experiments show MemDreamer achieves SOTA results across four mainstream benchmarks, narrowing the gap with human experts to only 3.7 points. It constrains the reasoning context window to merely 2% of full-context ingestion while delivering a 12.5 point absolute accuracy gain. Furthermore, statistical analysis uncovers a strong positive linear correlation between an VLM's performance on logic reasoning and long-video understanding benchmarks, establishing agentic capability scaling as a new paradigm for multimodal comprehension.