ADAS based on Deep learning for Tramway systems

The research illustrates a technique for real-time pedestrians, vehicles and cyclists detection and recognition along a tramway infrastructure in a complex urban environment by Computer Vision, deep learning approaches and the YOLOv3 algorithm. By the implementation of proposed technique in the Advanced Driver Assistance Systems (ADAS) can be increased road safety in terms of crash avoidance, crash severity mitigation and user’s protection during post incident phases. Experiments have been conducted in the tramway Line 2 “Borgonuovo –Notarbartolo” in the city of Palermo (Italy) in proximity to the tramway segment crossing a Three-arms roundabout having an external diameter of 24 m. The main results demonstrate that the method is able to very accurately detect and track the position of private vehicles, pedestrians and cyclists over or near the tramway track in front of the tram.