ISZOT: Extracting Zone-based Trajectory of Shopper for Marketing Analysis
- Yuan Li, Masanori Miyoshi, Shun'ichi Kaneko, Hironobu Fujiyoshi, Toshiki Etchuya, Hiroyuki Nara, Kotomi Watanabe
- 画像センシングシンポジウム, 2013
Download: PDF (English)
In marketing, sales are evaluated and improved by analyzing the purchasing behavior of shoppers, such as when, where and what they purchased. However, most analysis is based on POS (Point of Sale) data or on surveys carried out by human beings. This is not only costly and time consuming, but also lacks visual information on shoppers and their purchasing process. In order to overcome these problems, we propose a new and efficient surveillance system, ISZOT: Image based sensing system for analyzing Shoppers’ purchasing behavior based on ZOne Trajectories. This surveillance system uses images from surveillance cameras to extract a shopper’s trajectory in order to analyze their purchasing behavior for different marketing purposes. In this paper, ISZOT and its essential technologies will be introduced. In ISZOT, considering the accuracy and performance of image sensing process, a new human detection approach using a Smart Window Transform (hereafter SWT) and an edged-based classifier, and shoppers’ trajectories extraction approach is proposed. Experiments to evaluate the effectiveness of a store’s layout based on shoppers’ trajectories in an observed area are implemented with the cooperation of the Consumer Co-operative Sapporo, one of the largest chain stores in Hokkaido, Japan. The results show that our proposed system is capable and effective in analyzing shoppers’ purchasing behaviors for different marketing usages.