In this paper, we use convolutional siamese networks for face recognition. Face recognition is one of the non-contact biometrics and widely applied in many circumstances. As the advent of artificial intelligence era, the demands on accuracy and robustness get greater and greater at present. Due to its low accuracy and restrictions in many conditions, it’s hard for traditional methods to satisfy the demands. In this paper we propose a new network with convolutional siamese architecture for face recognition. Like some common face recognition systems, we use face detection to get the position of the face in an image. The deep learning methods are used to exploit the feature of the face. We finish the comparison of the detected face and the faces in the database. The experiments demonstrate a very good performance compared to other methods.