System for tolling

Abstract

A smart tolling system employs computer vision and machine learning techniques to automatically track and monitor vehicles across multiple video streams captured by a networked camera assembly. The system derives comprehensive vehicle trajectories by stitching together detections of each vehicle from the various video feeds using multi-object tracking algorithms. These trajectories enable correlating different events triggered by the vehicles at multiple roadside devices like cameras and radars. By mapping the trajectories in space and time, the system can associate seemingly fragmented events and synthesize complete vehicle profiles even when certain events may have missed capturing some vehicle information due to obstructions or other factors. This trajectory-based event correlation enhances tolling accuracy by minimizing revenue leakage from missed or incomplete vehicle data in congested traffic conditions.