Novel and Effective Approach for Multiview Biometric Object Detection using Deep Learning based Cutting Edge Techniques

Abstract

Currently, deep learning is experiencing rapid and significant growth in many fields. It offers solutions to a wide range of real-world problems. Additionally, its strong capacity for human authentication can help overcome challenges including human trafficking, border security, immigration control, etc. Many case studies failed to deal with crowdy places where human authentication is important. Addressing these challenges requires the development of a more robust and adaptable facial recognition model that can handle crowded and dynamic environments. So, in this paper, both RetinaFace and ArcFace (additive angular margin loss) are used to overcome the limitation. RetinaFace is employed for the detection of faces, while ArcFace is utilized to authenticate whether the detected faces are known or unknown. The main contribution of this work is to develop an application for personface recognition in crowdy places using RetinaFace and ArcFace. The model demonstrates strong performance, with an accuracy of 96% when tested in crowded places.