Calling Behavior Detection Of Port Truck Drivers Based On Deep Learning

Abstract

Due to the lack of a rigorous evaluation model in the traditional detection method of driver's illegal answering phone calls, it is difficult to meet the identification needs of truck drivers who answer the phone illegally in the port environment. A multi-feature fusion detection method based on deep learning is proposed. The method performs weighted fusion of the features of hand-held phone and speech to detect the calling behavior of port truck drivers. Mainly recognize faces through Retinaface training, and then extract the information of human key points through the PFLD_ CPN fusion key point detection model, and use YOLOv5 target detection to identify mobile phones; determine whether there is a phone call. The experimental results show that the method can effectively detect the behavior of answering calls in real-time and effectively in the self-collected monitoring screen data of port truck drivers.