n this intelligent transportation system project, vehicles in a highway traffic scene are detected and identified using video from a single camera in order to form an estimate of vehicular emissions. A novel algorithm will be developed to improve upon current computer vision and machine learning techniques for performing image segmentation in various weather, lighting, and traffic conditions, with a focus on improvement of segmentation in dense traffic conditions using a low resolution video source.
This algorithm will be used to classify vehicles by type (i.e. car, truck, motorcycle), identify vehicles by make and model, and track vehicle speed. The data captured by the algorithm will be matched with publicly available data from the Environmental Protection Agency (EPA) to perform a real-time estimation of emissions using the Motor Vehicle Emission Simulator (MOVES).
This project is partially sponsored by and/or in collaboration with the following: TranLive Initiative.