Intelligent Highway Speed Monitoring UAV System Based on Deep Learning

Abstract

At present, with economic development, traffic accidents occur frequently, and more than one-third of traffic accidents are caused by speeding. Current highway speed measurement devices have disadvantages such as high cost, inability to move, and low speed measurement accuracy. For these reasons, we propose an intelligent monitoring drone system for highway speed measurement based on deep learning. The system uses drones for monitoring, so the system is low-cost and flexible. First, the system's camera captures and recognizes vehicle movement and license plate information. Second, we propose algorithms based on the corners of the lane line speed measurement and the speed measurement based on the homography matrix to calculate the vehicle speed, and at the same time detect and recognize the license plate. Third, we propose to use ensemble learning methods to improve the accuracy of vehicle speed measurement, and finally obtain vehicle speed and speeding license plates. Experimental results prove that the proposed highway speed measurement UAV system can accurately identify and measure vehicle speed and record speeding license plates.