Performance Evaluation of Video-Based Face Recognition Approaches for Online Video Contextual Advertisement User-Oriented System

摘要

In this research, we propose the online video contextual advertisement user-oriented system. Our system is a combination of video-based face recognition using machine learning models from the camera with multimedia communications and networking streaming architecture using Meta-data structure to video data storage. The real images captured by the camera will be analyzed based on predefined set of conditions to determine the appropriate object classes. Based on the defined object class, the system will access the multimedia advertising contents database and automatically select and play the appropriate contents. We analyse existing face recognition in videos and age estimation from face images approaches. Our experiment was analyzed and evaluated in performance when we integrate analyze age from the face identification in order to select the optimal approach for our system.