Vehicle tracking in the network of traffic cameras is a must for future smart cities to facilitate, traffic management, autonomous driving, connecting cars informatically, and optimal route selection for all vehicles. It can also help to increase the safety of the elderly and visually people or children by predicting the dangerous situations. However, it is a very challenging task because often trees and infrastructure may block the view or sometimes larger vehicles obstruct the visibility of other vehicles, bikes, or pedestrians. Furthermore, large intersections make it hard to track vehicles’ movement vehicles. The variation of the light conditions, the viewing angle of cameras and weather conditions has a significant impact on the accuracy of the algorithms.
Sensifai and DAI-Labor joined forces to develop vehicle tracking across many traffic cameras of the city. This project which is supported by the European Data Incubator aims to incorporate the data collected by the network of traffic cameras in the self-driving journey of the vehicles. The pilot city is Berlin, where a network of traffic cameras and GPU servers are deployed.
In this project, we first develop a Vehicle detection system followed by vehicle tracking and multi-camera vehicle re-identification.