Абстрактный

Face Authentication Using Efficient Deep Convolutional Neural Network

Hayder M. Albehadili, Naz Islam

Recently face recognition, face verification becomes an interest scope of many researchers because they have been widely used in variety of real world applications. This work presents different strategies of constructing robust systems for face verification. An efficient deep Convolutional Neural network (CNN) is proposed to be used for face authentication. Although various methods are used to accomplish face verification, in this work, CNN is used instead because it has several merits over existing approaches. Exploring more robust CNN structures that can highly influence model accuracy is demonstrated in this work also. Although face authentication has received vast interest recently, there are very few works achieved using deep convolutional neural network. Thus, in this work, a robust model and efficient deep learning model is proposed to significantly enhance classification performance over existing works. The proposed models are evaluated using Labeled Faces in the Wild (LWF) dataset