Generating Analytic Insights on Human Behaviour using Image Processing

This paper proposes a method to track human figures in physical spaces and then utilizes this data to generate several data points such as footfall distribution, demographic analysis,heat maps as well as gender distribution. The proposed framework aims to establish this while utilizing minimum computational resources while remaining real time. It is often useful to have information such as what kind of people visit a certain place or what hour of the day experiences maximum activity, Such analysis can be used improve sales, manage huge number of people as well as predict future behaviour. The proposed framework is designed in a way such that it can take input streams from IP cameras and use that to generate relevant data points using open source tools such as OpenCV and raspberryPi.