Unifying Real-Time Multi-Vehicle Tracking and Categorization

Abstract

This paper addresses real-time automatic visual tracking and classification of a variable number of vehicles in traffic. This off-board surveillance device may cooperate with on-board Advanced Driver Assistance Systems (ADAS), extending its measurement range to the areas of the road that are not in the car sensors field-of-view (in a curve or an intersection). Tracking results also are useful for statistical trajectory analysis, devoted to understanding and improving user-user and user-infrastructure interactions. As a main contribution, this paper proposes to unify vehicle tracking and classification in a single processing step. This paper also addresses a vehicle anisotropic distance measurement based on the vehicle 3D geometric model. Real time tracking results are shown and discussed on road sequences involving various types of vehicles such as motorcycles, cars, light trucks and heavy trucks.