Fatigue Driving Detection Method And System Design Based On Door Control Loop Neural Network

In recent years,with the continuous increase in the number of cars,traffic accidents frequently occur,which has caused huge property losses to the country and individuals.At present,studies have shown that fatigue driving is one of the important causes of serious traffic accidents.The laws and regulations of countries around the world explicitly prohibit drivers from driving fatigued vehicles.Therefore,designing a system that can quickly detect fatigue driving behaviors and give drivers timely warnings has important social significance for improving traffic safety.The fatigue detection method based on driver’s visual characteristics has the advantages of low detection cost,real-time performance,and no interference to the driver,so it has been widely used in the field of automobile safety assisted driving.The detection of the driver’s face is a prerequisite for the fatigue detection method.However,factors such as occlusion of different types of glasses and changes in lighting during driving can have effects on the detection of driver’s face information.Aiming at the above problems,this paper first builds a reliable infrared camera image acquisition system,then combines the ideas of deep learning and human visual characteristics,and proposes a fatigue driving detection method based on a gated recurrent unit and fully convolutional network.The first step is to capture an image of the driver’s face with the help of an infrared camera acquisition system;the second step is to perform face detection and feature point location on the driver through a multi-task cascaded convolutional neural network(MTCNN),and obtain driver eye images based on the geometric position relationship of the key points on the face;the third step is to use the recognition algorithm of Convolutional Neural Networks(CNN)to identify the open and closed states of the extracted eye image,and output the results as complete facial state serialized data;the final step is to use the back-and-forth correlation of driver fatigue characteristics to transmit the state serialized data as a time series to a fully convolutional networks(FCN)with a gated recurrent units(GRU),and then determine whether the driver is fatigued.The experimental results show that the method proposed in this paper can accurately identify the open and closed state of the driver’s eyes in poor light conditions and when the driver wears different types of glasses.Compared with thefatigue parameter-based detection method,this method shows good fatigue detection performance on fatigue detection tasks,its accuracy rate reaches 99.12%,and it can realize the prediction of the driving state of the driver under experimental conditions.