Unconstrained face recognition using deep convolution neural network

Different methods have been proposed for face recognition during the past decades that differ essentially on how to determine discriminant facial features for better recognition. Recently, very deep neural networks achieved great success on general object recognition because of their potential in learning capability. This paper presents convolution neural network (CNN)-based architecture for face recognition in unconstrained environment. The proposed architecture is based on a standard architecture of residual network. The recognition performance shows that the proposed framework of CNN achieves the state-of-art performance on publicly available challenging datasets LFW, face94, face95, face96 and Grimace.