Research Progress on the Effectiveness of Deep Learning-Based Driver Fatigue Fusion Features

Abstract

In modern society, automobiles are more popular around the world, and people pay more attention to the safety of automobiles According to the information provided by the National Transportation Safety Administration in 2024, there are many traffic accidents caused by fatigue driving. This paper introduces two fatigue recognition methods, and studies the fusion of the two types of features. Aiming at the problem that the existing facial recognition methods are sensitive to environmental interference, this study systematically reviews the application of deep learning models such as YOLO, RNN and ResNet3D in this field from the perspective of facial feature analysis, and proposes targeted optimization strategies. For the problem of strong invasiveness and limited comfort in traditional physiological monitoring methods, this study innovatively constructs a multi-modal feature fusion framework. By collaboratively analyzing facial features and physiological signals, it can effectively balance the accuracy and comfort of monitoring while maintaining the advantages of non-contact detection. At the end of the article, the fusion mechanism of facial information recognition and physiological information recognition is further discussed. The fusion mechanism gives full play to the complementary advantages of the two types of information, and effectively reduces the adverse effects of single information recognition on driving behavior. As mentioned above, this paper analyzes a fatigue recognition method that combines facial information recognition with physiological information recognition. In other literatures, it is also mentioned that this scheme has the advantages of portability and accuracy compared with the traditional fatigue identification method.