基于自反思多模态理解的智能体式视频摘要
Agentic Video Summarization via Self-Reflecting Multimodal Understanding
把视频摘要从「特征提取+帧级重要性回归」的判别式范式，换成由 MLLM 驱动的「摘要—验证—反思」闭环智能体流程，视觉研究者可直接复用其自反思验证环节来提升关键帧选择的可解释性与自适应能力，而无需训练新网络。
原文依据 we propose AgenticVS--the first agentic workflow for video summarization that leverages multimodal large language models (MLLMs)
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论由大语言模型（LLM）驱动的 AI 智能体的兴起，通过使智能系统能够在多种任务中自主使用工具、推理与行动，改变了智能系统。尽管进展迅速，现有视频摘要方法主要聚焦于特征提取或帧级重要性回归，却缺乏真正基于智能体的智能所具备的自主推理、自我纠错与决策能力。为弥合这一差距，我们提出 AgenticVS——首个面向视频摘要的智能体式工作流，它利用多模态大语言模型（MLLM）以完全自主的方式完成「摘要—验证—反思」闭环。我们不为特征提取或回归设计新架构，而是利用 MLLM 的理解与反思推理能力，构建一个具备自反思工作流的自适应摘要框架。在 SumMe 与 TVSum 上的实验表明，我们的智能体式工作流优于最优方法，提升了可解释性与自适应性，并为基于智能体的多模态视频理解开辟了道路。
英文原文摘要 1129 字符
The rise of AI agents powered by large language models (LLMs) has transformed intelligent systems by enabling autonomous tool utilizing, reasoning, and action across diverse tasks. Despite this rapid progress, existing video summarization approaches primarily focus on feature extraction or frame-level importance regression but lack the autonomous reasoning, self-correction, and decision-making capabilities that define true agent-based intelligence. To bridge this gap, we propose AgenticVS--the first agentic workflow for video summarization that leverages multimodal large language models (MLLMs) to complete the summarization-verify-reflection loop in a fully autonomous manner. Rather than designing new architectures for feature extraction or regression, we exploit the understanding and reflective reasoning abilities of MLLMs to build an adaptive summarization framework with a self-reflecting workflow. Experiments on SumMe and TVSum demonstrate that our agentic workflow outperforms state-of-the-art methods, enhancing interpretability, adaptability, and paving the way for agent-based multimodal video understanding.