Intelligent Recognition of Vehicle Information in Surveillance Video

Abstract

In this paper, a method of real-time vehicle speed and vehicle type recognition is proposed. The vehicle type recognition is based on residual network to increase the convergence speed and improve the feature expression ability, and attach the centre loss to improve recognition accuracy of similar vehicles. The speed recognition is based on moving object detection. Experimental results show that the average error of vehicle speed is no more than 5%, and the average precision of vehicle type recognition is 85%, towards minibus and cars, the precision reaches 98.7%, which is superior to the traditional recognition method.