Face recognition using deep learning methods a review

Abstract

Face recognition is one of the most challenging field of image analysis and computer vision due to its wide practical applications in the areas of biometrics, information security, law enforcement and surveillance systems. It has been a topic of active research proposing solutions to several practical problems giving rise to the significant amount of research in recent times aimed at addressing the challenges of face recognition attributed to the following factors such as illumination, emotion, occlusion, facial expressions and poses, which great-ly affect the performance in achieving efficient and robust face recognition systems. In this field, many researchers adopted different techniques that solely rely on extracting handcrafted features to achieve better results. Recent development in deep learning and neural networks have made it possible to achieve promising results in numerous fields including pattern recognition and image processing. Deep learning methods boost up the learning process and facilitates the data creation task. Many algorithms have been developed to use deep learning architectures to get maximum result and achieve the state-of-the art accuracy. Some algorithms design their architectures from scratch and others fine-tuned the existing models to get maximum efficiency of generalization power. Algorithm complexity, data augmentation and loss minimization are the main concern of deep learning paradigms. We have reviewed these architectures in relation to algorithm complexity and experimental results on benchmark dataset. In this paper, we presented a literature survey of latest advances in researches on machine learning for face recognition and their experimental results on public databases.