Research on fatigue driving detection technology based on Dlib and micro-expression

Abstract

Driver fatigue is the most important factor contributing to the annual increase in traffic accidents and fatalities. Fatigue can affect driving performance due to lack of concentration and slower reaction time. Therefore, fatigue detection systems are crucial for safe driving. This article proposes a fatigue driving detection scheme based on the dlib library. Build a facial key point detection system based on the Dlib library. Real time calculation of eye and mouth data and head Euler angles using HOG-SVM algorithm, PnP algorithm, and fixed threshold. When running the YOLOV5 model in a PyTorch environment to recognize distracting behaviors, training is conducted on the sample and test sets for recognition, and the accuracy of the model is verified using the validation set for accurate recognition. Train smoking, answering and making phone calls, and drinking water behaviors to achieve distraction detection. Using convolutional neural networks to detect driver status and perform micro expression analysis. Fatigue assessment in multi feature situations of eye, mouth, and head posture and micro expression attention discrimination.