WCSE

Research on Vehicle Tracking and Collision Detection Method

Abstract— In order to quickly identify road vehicle collision accidents, a vehicle collision detection method based on machine vision is proposed. This method first uses the YOLO V5 deep neural model to achieve rapid recognition of vehicles; On this basis, Kalman filtering and Hungarian matching algorithm are introduced to achieve multi target tracking of vehicles; Finally, using the visual odometry calculation method, parameters such as vehicle speed, overlap, and trajectory deflection are obtained, and the occurrence of a collision is judged by whether the above parameters are abnormal. The test results show that compared with existing algorithms, this method has a significant improvement in recognition accuracy.

Index Terms— Collision detection, computer vision, multi-target tracing, visual odometry

Hong Weitian
Jianghan University, CHINA
Qi Tengfei
Jianghan University, CHINA
Liu Wenhao
Jianghan University, CHINA
Li Shaowei
Jianghan University, CHINA

[Download]

Cite: Hong Weitian, Qi Tengfei, Liu Wenhao, Li Shaowei, "Research on Vehicle Tracking and Collision Detection Method," 2024 The 14th International Workshop on Computer Science and Engineering (WCSE 2024), pp. 36-41, Phuket Island, Thailand, June 19-21, 2024.