Thinking with Cameras: Active Visual Reasoning via Dynamic Viewpoint Control for Surveillance Video Understanding
Paper • 2609.06475 • Published • 1
Thinking with Cameras: Active Visual Reasoning via Dynamic Viewpoint Control for Surveillance Video Understanding
Note High-quality training dataset for surveillance anomaly understanding
Note High-quality training dataset for camera viewpoint control
Note State-of-the-art open-source LVLM for real-world surveillance understanding