Exploring Image-Based Classification to Detect Vehicle Make and Model. Final Report

You are here: Homepage > Search

Exploring Image-Based Classification to Detect Vehicle Make and Model. Final Report

J. B. Flora / M. Alam / A. H. Yousef / K. M. Iftekharuddin

The goal of this work is to improve the understanding of the impact of carbon emissions caused by vehicular traffic on highway systems. In order to achieve this goal, this work obtains a novel pipeline for vehicle segmentation, tracking and classification by leveraging techniques in computer vision and machine learning using the existing Virginia Department of Transportation (VDOT) infrastructure on networked traffic cameras. This vehicle segmentation and classification data can be used to obtain a real-time estimation of carbon emissions. The VDOT traffic video is analyzed for vehicle detection and segmentation using an adaptive Gaussian mixture model algorithm. The segmented vehicles are tracked using speeded up reduced feature (SURF) methods. The morphological properties and histogram of oriented features are derived from the detected and segmented vehicles. Finally, vehicle classification is performed using a multiclass support vector machine classifier. The resulting classification scheme offers an average classification rate of 85% under good quality segmentation. This work constitutes the first step in estimating carbon emission for highway traffic. In the subsequent steps, we need to extract additional vehicle information (e.g., type, model), speed, and other relevant parameters and use this information for EPA's Motor Vehicle Emissions Simulator (MOVES) or similar tool to estimate the carbon emission.