Human Activity Recognition in Home Using Smartwatch and Wattmeter

-Human Activity Recognition Using Information of Human Motion, Indoor Location and Electric Power-

Abstract

Human activity recognition is one of the essential tasks for healthcare and plays a central role in the field of ubiquitous computing. Many works are found in the literature, however, they are prone to be privacy invasive, non-negligible for deployment and for sensing in practical real world setting. To avoid the above restrictions for practical use, we focus on smartwatch, and Bluetooth low energy beacons and smart power meter recently available as consumer products. Using these divices, motion, room-level location, and power consumption are aggregated in passive and continuous manner. In this paper, we propose a new framework using the above commercial devices for recognizing human activities including cooking, eating, cleaning where all the three sensory information is needed. We also deployed recognition systems to 3 households and collected over 800 minute data. The experimental result shows that all the three kinds of information are effective for recognizing our various daily activities.