Computer vision based people tracking for motivating behavior in public spaces

Abstract

In this work a system that automates the process of people counting to determine what effects "just-in-time" messages have on motivating behavior is described. The system is designed to permit automatic study of the impact of motivational messages on people's stair use. A projector presents a point-of-decision message to passers-by choosing between a set of stairs and an escalator while a computer vision algorithm counts each type of traffic. Preliminary results of the effects of messages displayed in a Boston area subway station are discussed. The system is designed to be easily moved to different locations with minimal change to the setup and algorithm. Results from an initial trail showed a 4.3% increase in stair usage (p < .001), demonstrating both the viability of the measurement technology and the potential of point-of-decision messaging to change behavior.