arXiv preprint / cs.CV

When and Where to Look: Adaptive Visual Evidence Scheduling for Efficient Long Video Understanding

作者：Ke Li, Jiayu Chen, Maoliang Li, Zihao Zheng, Hailong Zou, Hengyi Zhang, Xuanzhe Liu, Xiang Chen · 会议/期刊：arXiv preprint · 方向：cs.CV · 发布日期：2026-08-05

2026-08-05视觉感知
新标签打开 AlphaXiv新标签打开 Code
When and Where to Look: Adaptive Visual Evidence Scheduling for Efficient Long Video Understanding figure
AlphaXiv 中文概览（可滚动查看）