This project aims to design a Video Analytic Engine (VAE) architecture to support low-cost, real-time simultaneous facial analytics of video feeds from IP cameras at multiple locations. A deep learning model was customized for multiple demographic traits and facial feature extraction based on the parallel execution on GPU cores. The analytic core provides APIs to empower external recommendation engines to personalized product suggestions and advertisements for Smart Retailing. A Central Management System (CMS) is also designed to store, monitor, and visualize the customer metrics recognized.
