Traffic Related Air Pollution Prediction Using Video Footage and Deep Learning

Traffic-related air pollution is a major health burden in the United States. “Hyperlocal” pollution quantification is increasingly recognized as important for environmental sustainability and social equity, while traditional direct-measurement techniques do not scale well and are demanding in cost and labor. We are currently creating a system that uses traffic video footage and deep learning to predict measured traffic-related pollution concentrations around Urbana-Champaign.