中文基于AI视频监控的自杀风险评估：地铁站预防的可解释框架

该研究提出通过地铁监控视频联合分析乘客行为、空间上下文与时间动态，评估自杀风险，以支持早期干预。方法亮点在于融合人体运动感知与站台几何理解。实际意义：可提升高风险情境的自动识别能力，助力自杀预防。

arXiv:2605.22904v1 Announce Type: new Abstract: Understanding and monitoring human behavior in metro stations play an important role in supporting suicide prevention efforts, where early identification of high-risk situations can enable timely intervention. This requires assessing suicide risk from a surveillance video by jointly reasoning about the behavior of each passenger, his/her spatial context, and temporal dynamics. However, this assessment using videos captured by surveillance cameras is challenging, as it demands accurate perception of human motion, understanding of platform geometry