Suicide Risk Assessment from AI-powered Video Surveillance: An Interpretable Framework for Prevention in Metro Stations
A new framework for suicide risk assessment using AI-powered video surveillance in metro stations has been introduced. This framework aims to identify high-risk situations by analyzing passenger behavior and spatial context. The study achieved an 83.2% ROC-AUC on real surveillance data, highlighting the complexity of suicide risk assessment.
- ▪The framework assesses suicide risk by incorporating person tracking, activity recognition, and trajectory-driven risk heatmap modeling.
- ▪It formalizes the task of Suicide Risk Assessment (SRA) in metro stations, addressing challenges in human motion perception and behavioral cue aggregation.
- ▪This research opens new directions for interpretable AI systems aimed at social good.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Computer Vision and Pattern Recognition arXiv:2605.22904 (cs) [Submitted on 21 May 2026] Title:Suicide Risk Assessment from AI-powered Video Surveillance: An Interpretable Framework for Prevention in Metro Stations Authors:Safwen Naimi, Wassim Bouachir, Guillaume-Alexandre Bilodeau, Brian Mishara View a PDF of the paper titled Suicide Risk Assessment from AI-powered Video Surveillance: An Interpretable Framework for Prevention in Metro Stations, by Safwen Naimi and 3 other authors View PDF HTML (experimental) 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.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
