Eye State Detection Based on EAR and HOG PSO-Support Vector Machine

In order to detect the eye state of the dispatcher through the face image data, and then analyze and judge the fatigue state of the dispatcher, eye state detection method of the PSO-SVM support vector machine based on the EAR-HOG feature is proposed. Using the Retina-Face model to locate the key points of the face and the human eye, the single eye to be detected is obtained by the reference eye screening method, the EAR value and the HOG feature are calculated and extracted, and the SVM support vector machine optimized by the particle swarm algorithm is jointly input to classify the state of eye opening and closed. Using the self-made data set for verification, the experimental results show that the algorithm has high accuracy rate, while reasoning takes less time to meet the real-time requirements. It lays a technical foundation for further identification and classification of dispatcher fatigue states.