# MemDreamer：通过层次化图记忆与智能体检索机制解耦感知与推理的长视频理解

- 来源：HuggingFace Daily Papers（社区热门论文）
- 发布时间：2026-06-05 08:00
- AIHOT 分数：54
- AIHOT 链接：https://aihot.virxact.com/items/cmq7pt5at00jzslepjso94ysg
- 原文链接：https://arxiv.org/abs/2606.07512

## AI 摘要

MemDreamer 是一个即插即用框架，将长视频理解转化为智能体探索过程。它增量式处理视频，构建三层层次化图记忆（Hierarchical Graph Memory），用于语义抽象并捕获时空与因果关联。推理时，智能体通过观察-推理-行动循环进行工具增强的层次导航和节点搜索。在四个主流基准上，MemDreamer 达到 SOTA 效果，将人类专家差距缩小至 3.7 分，推理上下文窗口仅占全量输入的 2%，同时带来 12.5 个百分点的绝对准确率提升。统计分析发现，VLM 的逻辑推理能力与长视频理解性能呈强正线性相关，智能体能力扩展成为多模态理解新范式。

## 正文

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.