Real-time Face Recognition Under Different Environment

Abstract

The problems related to security and safety have motivated us to add some contribution to face recognition work by making it real-time process with more automation and robustness. As we know, fingerprint scanning is more manual as it needs human intervention. So Face is an unique identity of a person which is used by surveillance system. The aim of these surveillance systems is to provide an automatic interpretation of scenes by identifying correct person for human safety purposes. But due to lots of variation due to unconstrained environment and limitations face recognition becomes difficult task for real-time applications, where we will be considering few main problems like illumination, occlusion, head-pose and facial expressions will be solved in our work to improve real-time face recognition process. We will be using MTCNN and pre-trained Inception- ResNet model of CNN.