Detection of Abnormal Passenger Scenarios from Onboard Inside-Car Monitoring Cameras Using Object Detection Artificial Intelligence and Motion Analysis

In recent years, the increasing number of indiscriminate attacks on passengers aboard trains has prompted many railway operators to install onboard inside-car monitoring cameras for security purposes. Detecting abnormal passenger scenarios from videos captured through these cameras is critical for enhancing railway safety. In this paper, we present a method to detect abnormal passenger scenarios using onboard camera footage. To evaluate the proposed method, we conducted filming tests on an operational train, simulating abnormal scenarios involving a knife-wielding individual. Performance evaluation results demonstrate the effectiveness of the proposed method.