GroupShop: monitoring group shopping behavior in real world using mobile devices

In brick-and-mortar stores, customers often shop in small groups, which may be more common compared with shopping alone. Retailers in these physical stores, however, do not have effective methods to recognize who are potential buyers and how group members influence each other in purchase choices. This work aims to use mobile devices to help characterize these behaviors. Specifically, by utilizing the motion and acoustic sensors on mobile devices, we propose three features to describe different stages of group shopping and then identify potential buyers. Further, the influence behavior among group members is characterized based on the movements and interactions of customers. We have conducted real-life experiments with the recruited volunteers shopping in three environments. And the experimental results validate the effectiveness of our approaches, showing a high accuracy in group shopping behavior recognition.