Face fatigue detection method based on convolution neural network

Abstract

In order to monitor the fatigue state of the face, a state detection algorithm based on a variety of facial fatigue parameters is proposed. Firstly, convolution neural network is used to enhance the edge information of eyes and mouth for accurate positioning. Then, a rotation invariant LBP pyramid feature is used to describe the eyes, and a linear SVM classifier is trained to distinguish the open and closed state of eyes. The open and closed state of mouth is judged according to the open area and aspect ratio of mouth. At the same time, the vertical movement of eyes is counted Determine changes in head position. Finally, based on the state of eyes and mouth and the position of head, four fatigue parameters which can describe the state are calculated, and the final fatigue state is obtained by convolution neural network. The experimental results show that the detection and state discrimination algorithms have high accuracy.