Symphony: A Cognitively-Inspired Multi-Agent System for Long-Video Understanding

Haiyang Yan, Hongyun Zhou, Peng Xu, Xiaoxue Feng, Mengyi Liu, [1.5ex] Institute of Automation, Chinese Academy of Sciences · 2026

Read Symphony: A Cognitively-Inspired Multi-Agent System for Long-Video Understanding by Haiyang Yan, Hongyun Zhou, Peng Xu, Xiaoxue Feng, Mengyi Liu, [1.5ex] Institute of Automation, Chinese Academy of Sciences (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-17307.txt.

Description

Yan and co-authors present Symphony, a multi agent system for understanding long videos. Long video tasks pack dense information over long spans, and simple task splitting is not enough. Symphony mimics human cognition: it breaks the task into fine grained subtasks and adds a deep reasoning collaboration step with reflection. A vision based grounding approach scores how relevant each video segment is. The system reached state of the art results on four long video benchmarks. On LVBench it improved on the prior best by 5 percent.

Catalog record

CREATORHaiyang Yan, Hongyun Zhou, Peng Xu, Xiaoxue Feng, Mengyi Liu, [1.5ex] Institute of Automation, Chinese Academy of Sciences
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE52546
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.17307

Browse the full library · About the Tower of Babel