Video Analytics for Enhanced Safety and Efficiency in Ticketing Queues

Overcrowding is a prevalent issue in public transit systems, primarily due to the lack of effective real-time people counting measures. This results in problems such as theft, physical harm, disruptions, delays, and health & safety concerns, which ultimately compromise the safety of the transit system. To address this, we suggest a YOLOv8-based system for detecting overcrowding in real-time. This method utilizes predefined regions with specific thresholds; whenever the count of people in an area exceeds the threshold, an alert is triggered, allowing transit authorities to deploy countermeasures before overcrowding escalates. The paper explores the application of this region-threshold detection system and presents results based on real-world video data provided by a local public transit authority. Our findings show that the system performs effectively, with detection and alert accuracy ranging from 90 % in hightraffic scenarios to nearly 97 % in more moderate conditions, thus significantly enhancing efforts to mitigate overcrowding scenarios through continuous surveillance and data driven decision making.