Recognition of Remote and Small Intrusion Targets Around High-Speed Railway Based on Deep Learning Method

Abstract

With the rapid development of high-speed railway, its operation management and traffic safety are becoming more and more important. Foreign matter intrusion around high-speed railway needs to be timely prevented and identified. The current intrusion target recognition system still has the problems of high miss rate and false detection rate for the recognition of small intrusion targets beyond 100m. In this paper, aiming at this practical problem, combined with the deep learning target detection algorithm and the requirements of high-speed railway perimeter prevention and control, a system with better intrusion target recognition effect is designed to make up for the shortcomings of current high-speed railway perimeter video monitoring. Based on the actual video monitoring picture of high-speed railway, this experiment establishes a data set of far and small intrusion target recognition of high-speed railway perimeter, and establishes a complete evaluation standard system. Through the selection of algorithm model, it improves the ability of far and small intrusion target recognition of high-speed railway perimeter, and combines the algorithm to establish a far and small intrusion target recognition system of high-speed railway perimeter.