人物画像認識による来場者モニタリング

抄録

Recently, the needs of a profile measurement of customers in all market areas for the purpose of providing products and services that meet customer needs are increasing. This paper describes the development of a profile monitoring system of visitors using a video camera from the integration of human detection from upper body image and profile estimation from face image for the purpose of measurement. The upper body detection increase robustness of human detection with respect to the direction and overlap of a person. Moreover, a stable monitoring is realized by the synthetic estimation of visitors profile in multiple frames to track the person. In commercial facilities, the privacy of visitors will be required from the processing is performed in real-time recognition and non-saving video on the spot shooting. The implementation to commercial facilities which is achieved by performing the real-time processing with reduced cost recognition using appearance-based feature vectors is shown.