Face Recognition System in Unconstrained Environment through Convolution Neural Network

Compared with traditional machine learning methods, the learning characteristics of deep learning-based methods have received more and more attention in the past few years and have recently gained widespread popularity. This paper presents a novel method of face recognition in unconstrained environment based on deep residual ResNet50, standard convolution neural network architecture. The network is modified and fine-tuned for different facial recognition scenarios in an unconstrained environment. The goal of this Automatic Face Recognition (AFR) is to improving recognition performance with reducing the False Acceptance Rate (FAR) and False Rejection Rate (FRR). The proposed framework was trained and tested on the publicly available challenging datasets face94, face95, face96, Grimace and LFW and shows that it outperforms previous state-of-the-art methods without the need for additional resources.