Research Of Face Recognition Algorithm Based On Deep Neural Network

With the rapid development of computer technology,as well as the improvement of social information and network,all kinds of fields have put forward higher demand for fast and effective automatic authentication technology.Face recognition is widely used in the fields of finance,national security,justice,e-commerce and so on.Face recognition has become one of the hottest topics in the field of computer vision and pattern recognition.The development of deep learning and neural network has led to the development of face recognition technology.More and more deep learning models are used in face recognition and achieves good results.In this paper,a deep convolution neural network with multi convolution layers is used to extract features of faces.This model uses both the identification signal and the verification signal,and performs face verification by PCA and Joing-Bayesian algorithm.In this paper,the improvement of the pooling layer in the convolutional neural network is proposed.By replacing pooling operation with the convolution operation or the fully connected operation,it improves the ability of the model to acquire the information.At the same time,this paper proposes a model based on the location of face landmark,which can improve the accuracy of single model to more than 99%,also solve the time consuming problem of multi-model training.Using less than 700 thousand face images and input image size of 100,we achieved an average accuracy of 99.17%by the single model on the LFW dataset.