关键词: 驾驶安全识别；神经网络学习；智能驾驶安全；公共汽车

Abstract: This paper designs two programs, expression recognition and dangerous action recognition, such as the sudden physical illness of bus drivers, passengers grabbing the steering wheel, and the blankness of driving right transfer in the current L2 level intelligent driving. Among them, the expression recognition program is a face expression recognition program designed with Qt5 as the host computer using the Opencv computer vision library based on the Python language. First of all, the monitorability of relevant information is improved through the processing of input images, and the monitoring quality is improved, and then the expression model training and data acquisition are carried out, and finally the recognition function is realized. A LabVIEW-based dangerous action recognition program is also introduced. Through the video information transmitted by the camera, algorithms such as the inter-frame difference method and background learning are used to detect moving objects, so as to realize the real-time capture of foreign objects that may pose a threat to driving safety within the monitoring range. Experimental data show that the two programs can accurately and quickly capture the corresponding expressions and dangerous movements, which greatly improves the driving safety of buses while solving the above problems.

Key words: Driving safety identification; Neural network learning; Intelligent driving safety; Bus