Abnormal Behavior Analysis Strategy of Bus Drivers Based on Deep Learning
Shida Liu, Xuyun Wang, Li Wang, Xiaoping Zhang, Zhonghe He · 2021 IEEE 10th Data Driven Control and Learning Systems Conference (DDCLS) · 2021
Aiming at the bus driving safety problems caused by the abnormal behavior of the bus driver during the driving process, this paper proposes a deep learning-based analysis strategy for the abnormal behavior of the bus driver. The program defines the abnormal behaviors of bus drivers and categorizes them into behaviors such as smoking, drinking, and making phone calls. The YOLOv5 (You Only Look Once-Version 5) convolutional neural network algorithm is used as the core technique, and the abnormal behavior data of the drivers in the actual bus is used to produce the abnormal behavior data of the bus drivers. Collected and carried out automatic detection experiments to test the feasibility and effectiveness of drivers' abnormal behaviors. The experimental results show that the detection of abnormal behaviors of bus drivers is fast and accurate, the scheme is feasible and effective, and the detection effect can meet the application requirements.