W.X. Wang, B.G. Sun, R. Xia

Pages: 171-188

Abstract
Due to the poor accuracy, low efficiency and poor stability of fatigue driving detection of urban road at night, this paper proposes a fatigue driving detection of urban road at night based on multimodal information fusion. Firstly, the multi parameter extraction of fatigue driving state of driver's eyes, mouth and head is completed; Then, based on multimodal information fusion rules, the weighted average method is used to measure fatigue parameters and achieve classification of fatigue state levels. Finally, the fatigue detection model of the neural network is established, and the driver's fatigue detection is completed through SVM model classification. The experimental results show that this method can effectively realize accurate detection of fatigue driving of urban roads at night.
Keywords: fatigue driving at night; multi task learning; multi parameter extraction; multimodal information fusion; fatigue testing