Face Authentication Using Yolov8 and Facenet

The key aim behind this is to formulate a smart user Face Authentication System, that uses State-Of The Art Deep Learning methods to achieve secure, real time, contact less user authentication. There are many types of traditional authentication methodologies such as passwords which are prone to attacks and there is a need to choose a secured method. The system consists of two powerful models known as YOLOv8 for quick and accurate face detection, and FaceNet for extracting facial embeddings for identification. In the enrollment phase, the live video images are grabbed by the system and the faces are identified using an YOLOv8 detector, the facial embeddings are done with FaceNet. These embeddings are then saved into a NumPy array for subsequent references. Most facial data obtained in the authentication stage is newly collected facial data, embeddings are calculated by the new captured facial data compared to recorded embeddings by using cosine similarity. If similarity is above the threshold, the user is considered authenticated. The system is packaged as a Streamlit web application which has an user-friendly interface for enrollment and authentication. We focus on real-time performance, secure authentication, and great UX. Modular design allows for easy expansion into mobile integration, cloud deployment and multi-factor authentication. This work demonstrates the possibility of applying deep learning and computer vision into an efficient and practical face authentication system which strikes a fine line between privacy and usability in both personal and organization levels.