B0769
Title: Smartphone-based activity recognition using movelets
Authors: Emily Huang - Wake Forest University (United States) [presenting]
Abstract: Physical activity patterns can provide information about an individual's health profile. Traditionally, data on physical activity levels have been gathered through patient self-report in surveys. These data offer valuable firsthand accounts, but they can suffer from bias due to their subjectivity. In comparison, built-in sensors in the smartphone can measure data objectively and continuously, with less burden to the patient. There is a variety of data analysis approaches for smartphone-based human activity recognition. We applied the movelet method to classify the type of activity performed, based on smartphone accelerometer and gyroscope data. Our results show that this method has the advantages of being interpretable and transparent. A unique aspect of our movelet application is that it extracts information optimally from multiple sensors. Compared to single-sensor applications, our approach jointly incorporates the accelerometer and gyroscope sensors with the movelet method. Our findings show that combining data from the accelerometer and gyroscope can leverage their own distinct strengths, providing more accurate activity recognition than using each sensor alone.
