Classification of pedestrian behavior in a shopping mall based on LRF and camera observations

We analyze pedestrian behavior in a large shopping mall through observations using a laser range finder (LRF) and video cameras. The observed movements are classified into three categories, ‘going straight,’‘finding the way,’ and ‘walking around,’ based on each persons’ walking speed, variability of trajectory, stopping ratio, and head motions. Pedestrian behavior reflects the individual’s internal state (e.g., interests, preferences), and this analysis is intended to extract such information from the observed behavior. The obtained information is expected to be useful as marketing data for the the target facilities. It can also be useful in providing personalized on-line services to individuals corresponding to their interests and needs. In this paper, we analyze the observed properties from hand-labeled data and categorize a large-scale set of data based on the analysis. We also use our noncontact gaze tracking system to perform gaze analysis of pedestrians who look at a direction signboard in the shopping mall.