Facial Recognition System

Abstract

Real time facial recognition system is a turn-key solution deployed on a raspberry pi capable of identifying or verifying a person from a video frame. This would be achieved by comparing the facial features of the face detected from the video frame with the facial features of the face gallery present in the database. The main purpose of the system will be for security only but can be increasingly used for other purposes as-well. Initially, face recognition systems focused on still images. However, during the last years research on face recognition in image sequences has gained much attention, although nearly all systems apply still-image face recognition techniques to individual frames. In addition to its broader number of applications, video-based face recognition provides several advantages over still image based face recognition. Our system works by first obtaining the input through a Raspberry Pi Cam in the form of a live video feed. The image will be extracted frame by frame from the video for further processing. Face detection will be performed on each frame. The technique that we will be using for face detection is the Voila and Jones face detection algorithm. After the face detection step, the face recognition step will be performed in two modules. The first is the feature extraction for which we are using the Histogram of Oriented Gradient (HOG) features and Local Binary Pattern (LBP) features. For the feature classification module we will be using the Multi Class Support Vector Machines (SVM) for features classification. The final result will be obtained from the classifier and will be labeled corresponding to the face in the model.