arXiv preprint / cs.CV

CoEvoWhen: Policy-Tool Coevolution for Ultra-Long Video Temporal Grounding

作者：Yiduo Jia, Muzhi Zhu, Jinchuan Shi, Hao Zhong, Yuling Xi, Ke Liu, Hao Chen · 单位：Zhejiang University, State Key Lab of CAD & CG https://aim-uofa.github.io/CoEvoWhen · 会议/期刊：arXiv preprint · 方向：cs.CV · 发布日期：2026-10-01

2026-10-01视觉语言视觉感知强化学习
新标签打开 AlphaXiv新标签打开 Project
CoEvoWhen: Policy-Tool Coevolution for Ultra-Long Video Temporal Grounding figure
AlphaXiv 中文概览（可滚动查看）