A Driver Fatigue Detection Framework with Convolutional Neural Network and Long Short-Term Memory Network

A Driver Fatigue Detection Framework with Convolutional Neural Network and Long Short-Term Memory Network

연구 분야: Networking

학회: International Conference on Applied Intelligence and Informatics

Fatigue driving is a significant cause of traffic accidents and injuries. To reduce this risk, a compact and effective method for detecting driver fatigue using deep learning techniques is proposed. The method combines a 2D-Convolutional Neural Network (CNN) and a Long Short-Term Memory network (LSTM), enabling the spatial and temporal dynamics of driver behavior to be effectively captured. To enhance the model’s robustness and generalization ability, several types of data augmentation are also applied. The proposed method is evaluated on the publicly available YawDD dataset, achieving an accuracy of 95% in distinguishing between fatigued and alert driving behavior. The effectiveness of 2D-CNN-LSTM networks for driver fatigue detection is demonstrated, with potential practical applications in the automotive industry highlighted. The analysis of the results reveals that the proposed method outperforms some existing methods in the literature, although there is still room for improvement. The strengths and weaknesses of the method are discussed, along with potential directions for future research.