Automated System for Preventing Suicides by Train

Suicides in public have devastating consequences for families, the local authorities dealing with the aftermath of such incidents, and witnesses. Subways, metros, and trains are one of the most accessible modes of public transport and are emerging as one of the most widespread forms of vehicular suicide. Although measures like placing more guards on platforms and using glass barriers to prevent direct access to the railway tracks are being used in multiple railway stations across the world, their implementation in old stations is inefficient and expensive. This paper proposes a method that uses Image Analysis and other Computer Vision techniques on real-time closed-circuit television (CCTV) feed to detect behaviors that indicate suicidal intentions and, if detected, notify the rail personnel for appropriate countermeasures. The proposed method judges each individual based on eleven parameters, considered as indicators of suicidal intentions. Experimental observations on limited datasets, even with low video quality, have demonstrated the high precision of our system in predicting potential suicides by train.