Smartphone Based Activity Recognition System
Sen Zhang · OhioLink ETD Center (Ohio Library and Information Network) · 2012
Nowadays, obesity and overweight are epidemic worldwide.Obese people are at risk for diseases such as heart disease, type 2 diabetes, stroke, and certain types of cancer.Finding out an effective treatment for obesity becomes a major challenge for researchers.Physical activity plays a vital role in treating obesity, as the energy consumed during such activity helps "burn off" excess body fat.With the rapid development of smartphone technology, we realized that prolific mobile phones can potentially monitor physical activity thanks to their motion sensors, e.g., accelerometers, orientation sensors, and gyroscopes.In this paper, we design and implement an activity recognition system using mobile phones.Unlike prior work, our system uses acceleration, orientation, and gyroscope data for on-line activity recognition.Our system samples phone sensor data at 5 Hz and uploads them to a central server for processing.Our preliminary experiments indicate our system's potential for activity recognition.