Research on Face Recognition Algorithm Based on Deep Learning

In this paper, a feasible face recognition algorithm based on CNN method and TensorFlow deep learning framework is proposed for face multi-pose and occlusion recognition problems. Firstly, a face detection model was established. Under the TensorFlow learning framework, the MTCNN face detection model was established and trained with Wider Face data set. Secondly, the face recognition model was established, the Facenet algorithm model was established, the CASIA- Webface dataset was used for training, and the LFW was used for model testing. The test results on LFW data set show that the face recognition accuracy of the proposed algorithm can reach 98.78^%.