情報処理学会 第81回全国大会講演要旨

Customer's product selection behavior in retail environments provides valuable information toward improving marketing effectiveness. In such environments, a customer may consider buying a product by taking it off the shelf, analyze it, but end up not purchasing it and return it back to the shelf. In this research, such behavior is detected with the top-view camera’s images. We detect and collect data of customer’s hand’s position and state of holding products. Then a behavior history is created according to several combinations of these data. The history is classified to output a sequence of understandable customer behavior. This sequence reveals customer's selection process which is helpful for further analysis of customer behavior.