Face-Pose Estimation for Learning Systems

Face recognition is the most known technique that used in our daily lives. In computer vision applications there are numerous reasons for the growing interest in automated face recognition systems. Nevertheless, face-pose variation stills the main challenge in face recognition systems. This paper suggests a face-pose recognition system that predicts and classifies face-pose angles for students during studying. The bounding-box method for face detection is used and a high-dimensional vector-space of faces using HOG features is adopted. A mixture of linear regression technique is applied to determine how to map high-dimensional vector-space onto bounding-box shifts and both face-pose parameters. Then eight classifiers are applied to recognize face-pose angles of students in learning environment. The execution of the suggested system is evaluated and checked the mean absolute error and standard deviation. Also, the accuracy and misclassification rate are used as measures of performance of the introduced system. The proposed method showed good performance in all of these measures.