Shopper behaviour analysis based on 3D situation awareness information

Abstract

The customer behaviour understanding is of major importance to brick and mortar retail struggling to keep their market share and competing with online retail. In this paper, we propose a customer behaviour tracking solution based on 3D data. We can cover large areas using numerous inexpensive networked 3D sensors for monitoring and tracking people and we have adopted an adaptive background model in order to be able to react to changes in the store environment. Experiments with people tracking and analysis of the trajectories in a department store show that use of inexpensive 3D sensors and lightweight computation allows classifying shopping behaviour into three classes (passersby, decisive customers, exploratory customers) with 80% accuracy.