Low-Power Traffic Surveillance using Multiple RGB and Event Cameras: A Survey

Smart city intersections may support connected automated vehicles in the future and are the core of intelligent traffic management and control. In this work, we consider the traffic surveillance task in multi-camera and event-camera setups at road intersections with a focus on efficient edge computation. We give an overview over different aspects of such systems from the literature, including relevant recognition tasks, architectures, benchmark datasets and algorithms with a focus on their properties like real-time capability and energy consumption where possible.