Vehicle trajectory data

           The vehicle trajectory data, as a type of traffic microdata, has attracted more and more researchers' attention. It can be used for various traffic researches, e.g. car following behavior, vehicle lane changing behavior, and road safety analysis etc.

           A vehicle detection and tracking system based on CCTV cameras is developed for collecting trajectory data. A deep learning based method, called convolutional neural network(CNN), is used to achieve high precision vehicle detection and tracking. This system can identify multi-type vehicles, e.g. car, vans, truck, and motorcycle. Pedestrian detection and tracking is also supported. This system is supposed to build a big-data system for traffic research.