Deep Tracking and 3D Monitoring in Traffic Analysis

Abstract

As cities grow and self-driving cars become more common, the use of computer vision in intelligent traffic systems and traffic monitoring becomes increasingly important. Advanced traffic management systems today leverage video surveillance infrastructure with deep neural networks. This study demonstrates a practical and efficient platform for real-time traffic monitoring using a single CCTV traffic camera. The platform includes 3D vehicle and pedestrian detection, speed detection, trajectory estimation, congestion detection, and vehicle-pedestrian interaction monitoring.