Design and implementation of a targeted advertising system based on an audience recognition scheme for Smart TV

Abstract

This paper1 suggests a targeted advertising system using an audience recognition technique suitable for a Smart TV environment. The proposed system detects different viewers in front of a TV and provides different advertisements depending on the detection results. Audience recognition consists of procedures for detecting the face area and features, as well as an identification procedure; in addition, based on the results of the recognition, the viewing behaviors are extracted. In regard to the extraction of the face area and the facial features, algorithms for a fast and precisely trace even when the face is rotated, and for measuring the rotating angle of the face and the open-close status of the eyes, are described. All recognition procedures were designed by considering the viewing distance and changes in lighting. In addition, a practical targeted advertising service and various other measurement indexes related to advertising effectiveness, along with an example application used to control the equipment and content by applying identification information of the recognized audience and their viewing behavior information, are represented.