Head orientation estimation using neural network

In this paper; we propose neural-network based schemes to solve the head orientation estimation (HOE) problem. Faces are detected using the Ycbcr skin detection method, and then we labeled the detected faces k key points manually that present different orientations, the coordinates and local textures of the k key points are obtained to compose the input feature vectors of the neural networks. By training 1300 data sets, the results show that the neural network based method can estimate head orientation at the correct rate of 90%.