Face Recognition of Live-Feed Imagery Using the Viola-Jones Algorithm

This paper proposes the design, execution and testing of a live facial detection and recognition system. The system is intended to detect a face from live footage as an implementation of the famous and successful Viola-Jones (VJ) algorithm and recognize the face via an implementation based on Python libraries. This implementation uses pictures of a handful of people as training data and attempts to recognize the same people when they appear in front of the web camera. Testing of this system confirmed the ability of the system to fulfil this function satisfactorily. Within a limited margin of error, this system strives to satisfy the requirements of speed and resilience. Face recognition is the primary focus of this system, which makes use of Python libraries to perform the tried-and-true VJ method of detecting faces in real-time video. Using just a tiny subset of people's faces as training data, this method attempts to recognize them again when they pose for a webcam. Testing confirmed that this functionality could be successfully carried out by the system.