Research On Fatigue Driving Detection Technology Based On Facial Features

With the rapid development of national economy and the remarkable improvement of people’s living standards,more and more people choose to buy private cars.The increase in the number of cars leads to frequent traffic accidents,causing a lot of property damage and personal injury.Among the many causes of traffic accidents,fatigue driving accounts for a large proportion.Therefore,it is important to study an effective driver fatigue detection method.In this paper,a large number of related literatures are reviewed,three commonly used fatigue driving detection methods are compared,and the fatigue driving detection method based on face features is selected,which does not interfere with the driver and has high detection accuracy.There are limitations and low accuracy in judging fatigue degree based on a single fatigue feature,and when the driver is in a fatigue state,multiple fatigue states of the face will change,such as frequent blinking,yawning,drowsy nodding or head tilting,etc.Therefore,this paper uses multiple fatigue features fusion for driver fatigue determination,and proposes a fatigue driving detection method that is non-contact and capable of real-time detection to improve the accuracy of fatigue determination and reduce the probability of misclassification.This paper firstly uses the improved target detection network YOLOX to detect the driver’s face region.YOLOX has the advantages of fast detection speed and high accuracy,which can well meet the comprehensive requirements of accuracy and speed for face detection in this paper.Then,the improved PFLD face key points detection model is used for face key points location and head pose estimation,which can not only meet the requirements of fast key points detection and high accuracy,but also use head pose parameters to calculate head-related feature values,which reduces the amount of calculation.Finally,a multi-feature fatigue discrimination model is used to determine the fatigue level of the driver in real time and perform fatigue warning.Experiments show that the fatigue driving detection method proposed in this paper has an accuracy of more than 97%,and the real-time FPS can reach 46frames/second,which can effectively perform real-time fatigue driving detection,and the performance of the method is better than most other fatigue driving detection methods.This paper uses Python language to implement a specific detection algorithm,and combines Py Qt5 to complete the design of the detection work interface,and through the image and data real-time visual display of fatigue detection results.The results show that the detection method proposed in this paper can meet the needs of fatigue driving detection.