Design of Real-Time Vision-Based Face Verification and Tracking Via Pan Tilt Camera

Computer vision and autonomous systems are sub-fields of artificial intelligence. Vision-based applications are significantly applied in surveillance, healthcare, and assistance fields. Pan Tilt Camera is assembled and implemented in the proposed system. The face verification stage in Pan Tilt-based face tracking system is processed using One-Shot Encoding models. The study indicates the implementation of two encoding models, Histogram of Oriented Gradient (HOG) and Convolutional Neural Network (CNN) encoding models. The locomotion algorithm is based on a sequence of arithmetical equations. The CNN encoding model performance shows more accurate verification values, whereas HOG encoding model performance shows less accurate verification values. The HOG encoding model-based tracking performance shows highly accurate, streamlined, and rapid locomotion- behavior, whereas the CNN encoding model-based tracking performance shows less accurate, inefficient, and slow locomotion-behavior.