High-quality traffic data are crucial for infrastructure planning, system operations and performance measurement, safety considerations, maintenance activities, and informed analysis and decision-making. That’s why state departments of transportation need comprehensive and cost-effective traffic sensing and data collection systems. The primary goal of this research is to develop machine-learning-based detection algorithms, software, and a mobile hardware system that can utilize existing surveillance video cameras to accurately collect critical traffic information that traditional traffic sensors often cannot capture. This information includes vehicle volumes based on FHWA’s 13-bin classification system, speeds, and road surface conditions. The machine learning process will allow the unit to be trained with real data. The major advantage of this new system will be its ability to collect short-duration count data where geometry and volumes pose safety risks to field staff.
Principal Investigator: Yinhai Wang, Department of Civil and Environmental Engineering, UW
Sponsor: WSDOT
WSDOT Technical Monitor: Natarajan Janarthanan
WSDOT Project Manager: Doug Brodin
Scheduled completion: December 2025