Roussou S., Mirogianni N., Yannis G., “Investigation of Non-Compliant Pedestrian Crossings at Signalized Intersections Using Computer Vision Techniques”, Proceedings of the Transport Research Arena TRA 2026 Conference, Budapest, 18-21 May 2026.

Pedestrian safety at signalized intersections is an important issue in urban areas, especially in areas with high population density, like the city of Athens. The aim of this study is to examine pedestrian non-compliant crossing behavior at a signalized intersection through video data collection via both manual and automated approaches. The automated approach is carried out through a computer vision system that includes object detection, trajectory detection, homography transformation, and Time-to-Collision (TTC) estimation. The aim of this study is to compare the results obtained through manual and automated approaches and examine non-compliant pedestrian crossing behavior at signalized intersections. The study also examines the characteristics of non-compliant pedestrian crossing behavior and pedestrian behavior at signalized intersections. From the results, it is concluded that the performance of automated detection is influenced by factors like occlusion and signal visibility, and pedestrian non-compliant behavior is influenced by factors like signal timing, waiting time, and traffic. The study helps understand pedestrian non-compliant behavior and also demonstrates the potential for analyzing pedestrian behavior at signalized intersections through automated approaches like computer vision.