Activity monitoring using smart glasses: Exploring the feasibility of pedometry on head mounted displays

Abstract

Fitness tracking, fall detection, indoor navigation, and visual aid applications for smart glasses are rapidly emerging. The performance of these applications heavily relies on the accuracy of step detection, which has rarely been studied for smart glasses. In this paper, we develop an accelerometer-based algorithm for step calculation on smart glasses. Designed based on a salience-analysis approach, the algorithm provides a highly accurate step calculation. An activity monitoring application for a commercial Android-based smart glasses (Vuzix M100) is designed and realized for algorithm evaluation. Experimental results from 10 participants wearing the smart glasses running our application achieved average step detection error of 2.6% demonstrating the feasibility of our salience-based algorithm for performing pedometry on smart glasses.