Publication Date: 2024/01/03
Abstract: Face recognition is a concept of the safest way of logging on; it entails that our facial images are acquired, detected, and subsequently authenticated by the particular interface. In this present digital generation, safe authentication of the interfaces is the primary cautionary aspect that should be maintained, and this model suggests a secure and strong authentication system. This paper recommends a face recognition login interface that involves deep learning models to provide a strong and secure authentication mechanism. It involves the extraction of facial images, proposes a solution to enhance accuracy and trustworthiness, and presents a weighty improvement over the traditional username-password login method. This gives us a user-friendly login experience along with the highest level of security. These facial authentication models are being used in numerous fields, ranging from security, healthcare, marketing, retail, public events, payments, door unlocking and video monitoring systems, user authentication on devices, etc. It is also useful for multi-class classification problems. This paper includes face recognition techniques from convolutional neural networks (CNN) and transformer models like ViT (vision transformer), VGG16, RestNet50, Inception V3, and EfficientNetB0. It proposes that the best model will be deployed using Streamlit.
Keywords: Secured Authentication System, Facial Recognition, Deep Learning, Vision Transformer, VGG16, Image Classification, Streamlit.
DOI: https://doi.org/10.5281/zenodo.10453298
PDF: https://ijirst.demo4.arinfotech.co/assets/upload/files/IJISRT23DEC1536.pdf
REFERENCES