AI 中文总结
该研究提出Video-DR智能体框架，通过解耦感知-探索流水线与分阶段工具解锁解决现有多模态模型的模态偏差和知识泄漏问题，构建基准并在视频问答任务上取得SOTA性能。
AI 中文摘要
我们提出Video-DeepResearch（Video-DR），将多模态智能体从静态图像扩展到连续视频流，该场景需要密集的时空定位能力结合开放网络探索。初步评估显示当前模型存在两个关键瓶颈：（1）模态偏差，即智能体绕过视觉工具而倾向于文本搜索；（2）参数知识泄漏，即模型依赖内部记忆而非真正的工具增强执行。为应对这些挑战，我们提出Video-DR，其具有解耦的感知-探索流水线，采用分阶段工具解锁机制，强制在网络检索前进行详尽的跨帧视觉定位。我们的框架采用两阶段训练方案：监督微调后接分组相对策略优化（GRPO），实现突破模仿学习上限的自主探索。此外，我们构建了Video-DR-Bench，这是一个包含200个复杂多跳视觉问答实例的人机协作基准。实验结果表明，我们的Video-DeepResearch-35B-A3B取得了64.0%的平均准确率，超越专有模型Claude-4.5-Sonnet（59.0%）5.0个百分点，且显著优于GPT-5（52.5%）和Gemini 2.5 Pro（57.5%）；30B-A3B变体达到59.3%，与Claude-4.5-Sonnet具有竞争力，证明了我们的训练范式在紧凑规模下的有效性。代码：this https URL。
英文摘要
We introduce Video-DeepResearch (Video-DR), extending multimodal agents from static images to continuous video streams, a setting that demands dense spatiotemporal grounding coupled with open-web exploration. Preliminary evaluations reveal two critical bottlenecks in current models: (1) modality bias, where agents bypass visual tools in favor of textual search, and (2) parametric knowledge leakage, where models rely on internal memory rather than genuine tool-augmented execution. To address these challenges, we propose Video-DR, featuring a decoupled perception-exploration pipeline with stage-wise tool unlocking that compels exhaustive cross-frame visual grounding prior to web retrieval. Our framework adopts a two-stage training recipe: supervised fine-tuning followed by Group Relative Policy Optimization (GRPO), enabling autonomous exploration that breaks the imitation-learning ceiling. Furthermore, we curate Video-DR-Bench, a human-AI collaborative benchmark comprising 200 complex, multi-hop VQA instances. Empirical results demonstrate that our Video-DeepResearch-35B-A3B establishes a new state-of-the-art of 64.0% average accuracy, surpassing proprietary Claude-4.5-Sonnet (59.0%) by 5.0 points and significantly outperforming GPT-5 (52.5%) and Gemini 2.5 Pro (57.5%). The 30B-A3B variant achieves 59.3%, competitive with Claude-4.5-Sonnet and demonstrating the effectiveness of our training paradigm even at compact scale. Code: https://github.com/Osilly/Vision-DeepResearch.
