A Fast and Efficient Face Pose Estimation Algorithm for Mobile Device

摘要

This paper proposes a method for face pose estimation from a single image. Face pose estimation is mainly used to estimate the 3D Euler Angle of the input face region image and is widely applied in human-computer interaction, video surveillance and other fields. For the purpose of improving the speed and accuracy of face pose estimation, multifarious operators or modules are utilized to build an efficient convolution neural network. Moreover, a variety of methods for data enhancement are used while training the CNN. Experiments show that our proposed method has higher speed and accuracy compared with previous face pose estimation methods.