Abstract :

This project implements a real-time face-recognition-based authentication system using computer vision and deep learning. The system captures live video frames through a webcam and uses the face_recognition library to extract facial embeddings. A pretrained facial feature encoding is generated from a reference image and compared against faces detected in the camera feed. If the similarity score matches, the user is authenticated; otherwise, access is denied. The model leverages Histogram of Oriented Gradients (HOG) and deep neural network–based facial encoding for robust identification. The solution ensures fast processing, high accuracy, and contactless authentication suitable for access control, automated attendance, IoT devices, and smart security applications.