Event-Based Sensing for Improved Traffic Detection and Tracking in Intelligent Transport Systems Toward Sustainable Mobility

This paper presents a pipeline that utilizes a TorchScript model to implement event-based detection in challenging traffic scenarios, aligning with the theme of Advances in Intelligent Transport Systems (ITS) for Sustainable Mobility. The pipeline incorporates intuitive interfaces for visualizing event streams in both 2D and 3D, enhancing the understanding of traffic dynamics. It also integrates adaptive rate control and optical flow estimation techniques to improve detection and tracking capabilities. The model demonstrates promising results in accurately detecting and tracking vehicles and pedestrians. To validate the model’s performance, we analyze the spatio-temporal distribution of events using histogram difference computation and exponential-decay time surface analysis, which provide valuable visual insights. By effectively utilizing event-based sensing and incorporating innovative techniques, this research contributes to the advancement of vision-based traffic monitoring for sustainable mobility within the field of Intelligent Transport Systems.