Research on Intrusion Detection Method of Key Metro Areas based on YOLOv3
Meijie Li, Zhengyu Xie, Yong Qin, Yingqing Mai · 2020
The metro station is one of the public places with the largest flow of people. It is very important for passengers and station staff to detect whether there is any intrusion in its key area. There are many defects and deficiencies in the past human detection methods. With the popularization and development of computers and the rise of deep learning methods, a large number of visual detection methods have emerged and gradually occupy an important position in the field of video detection. In this paper, we propose a regional intrusion detection method, based on the YOLOv3 target detection method of the deep learning method. The experiment is carried out on the monitoring video of a metro station to test the reliability of our method. The final experimental results show that our detection method has the advantages of fast detection speed, high detection precision, and can complete the detection task as well.