Despite significant recent achievements in the field of face recognition, implementing face sketch recognition effectively presents serious challenges to existing methods. In this paper, we propose a novel face sketch recognition approach based on FaceNet which can directly map face image to a feature vector. Instead of matching the face sketch images to photo images directly, we firstly transfer the photo images in gallery into sketch style to reduce the great texture discrepancy between face sketches and photos. Extensive experiments on public datasets demonstrate the superiority of the proposed method.