arXiv 2506.090792025-09-26cs.CVcs.AI

VidBridge-R1：通过中间代理任务桥接基于强化学习的视频理解模型中的问答与字幕生成

VidBridge-R1: Bridging QA and Captioning for RL-based Video Understanding Models with Intermediate Proxy Tasks

New Laboratory of Pattern Recognition (NLPR), Institute of Automation, Chinese Academy of Sciences (CASIA)（中国科学院自动化研究所模式识别实验室）；School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院）；Kling Team, Kuaishou Technology（快手科技 Kling 团队）；Peking University（北京大学）；Nanjing University（南京大学）

Xinlong Chen, Yuanxing Zhang, Yushuo Guan, Weihong Lin, Zekun Wang, Bohan Zeng, Yang Shi, Sihan Yang, Qiang Liu, Pengfei Wan, Liang Wang, Tieniu Tan

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对强化学习视频模型在问答与字幕任务上难以兼顾的问题，提出基于DarkEventInfer和MixVidQA两个中间代理任务的训练框架VidBridge-R1，使单一模型在两类任务上均获显著性能提升。

摘要：

由强化学习增强的“先推理后响应”范式在推进多模态大语言模型方面展现出巨大潜力。然而，其应用于视频领域时，导致模型出现专业化现象：要么擅长问答任务，要么擅长字幕生成任务，但难以同时精通两者。简单地将这些任务的奖励信号进行组合会导致相互性能下降，我们将其归因于这些任务在本质上相互对立所引发的冲突。为解决这一挑战，我们提出了一种基于两个中间代理任务的新型训练框架：DarkEventInfer，该任务向模型展示带有掩蔽事件片段的视频，要求模型根据上下文视频线索推断被遮挡的内容；以及MixVidQA，该任务向模型展示由两个不同片段组成的交错视频序列，挑战模型在忽略其中一个片段的同时，对另一个片段进行隔离和推理。这些代理任务迫使模型同时发展整体性、发散性理解能力以及精确性、收敛性推理能力。基于该框架，我们提出了VidBridge-R1，这是首个能够有效弥合范式冲突的多功能视频推理模型。大量实验表明，VidBridge-R1在单个模型中同时实现了问答和字幕生成任务的显著性能提升，证明了我们的方法在培养更具泛化性和更强大的视频理解模型方面的有效性。

英文摘要：

The "Reason-Then-Respond" paradigm, enhanced by Reinforcement Learning, has shown great promise in advancing Multimodal Large Language Models. However, its application to the video domain has led to specialized models that excel at either question answering (QA) or captioning tasks, but struggle to master both. Naively combining reward signals from these tasks results in mutual performance degradation, which we attribute to a conflict between their opposing task natures. To address this challenge, we propose a novel training framework built upon two intermediate proxy tasks: DarkEventInfer, which presents videos with masked event segments, requiring models to infer the obscured content based on contextual video cues; and MixVidQA, which presents interleaved video sequences composed of two distinct clips, challenging models to isolate and reason about one while disregarding the other. These proxy tasks compel the model to simultaneously develop both holistic, divergent understanding and precise, convergent reasoning capabilities. Embodying this framework, we present VidBridge-R1, the first versatile video reasoning model that effectively bridges the paradigm conflict. Extensive experiments show that VidBridge-R1 achieves significant performance gains on both QA and captioning within one model, demonstrating the efficacy of our approach in fostering more generalizable and powerful video understanding models.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。