Customer classification method for multiple pedestrians using pressure sensors

Abstract

This paper aims to customer classification age groups by walking data using pressure sensor array. Techniques of the age groups estimation in many retail businesses (for example, convenience stores, supermarkets, shopping malls, etc.) are marketable. Because those techniques are applicable for market research and customer analysis. There are many researches of the age estimation using face images, walking silhouette data, etc. However, there are some problems too. One of problem is that the estimation classes are a few. Moreover, many age estimation systems use some video cameras. Therefore, these systems may invade surveyed person's privacy by taking one's face images. Obviously the pressure sensor does not record one's face images. In this study, this fact is one of merit in using the pressure sensors. The pressure sensor array gets feature quantity including center of gravity, pressure value, area of foot, etc. In this study, our system classifies surveyed persons to 6 age groups as each decade. Here, surveyed persons are 20s-70s. Average estimation accuracy of all age groups is 82.67%. The highest estimation accuracy is 89.33% at 20s and 60s.