新闻
When and Where to Look: Adaptive Visual Evidence Scheduling for Efficient Long Video Understanding
arXiv cs.AI · 发布于 · 阅读约3分钟
30秒读懂
- 发生了什么
- EcoFrame framework adaptively selects video frames for vision-language models using entropy-based budget scheduling and attention-guided search instead of static frame selection.
- 为何重要
- Matters for engineers building long-video understanding systems who need faster inference without sacrificing accuracy on benchmarks like Video-MME and LongVideoBench.
- 注意
- Training-free approach requires the VLM to already be deployed; unclear how well entropy signals generalize across different model architectures and video domains.
收听本摘要
- agent
- language model
- reasoning
- rag
- inference
The Agent Architect
每周一个模式、一个权衡、一个生产事故案例。为构建智能体系统的人准备的每周简报。
每周一封邮件,一键退订。您的地址仅用于发送简报。