Human Daily Activity and Fall Recognition Using a Smartphone’s Acceleration Sensor

Abstract

As one of the fastest spreading technologies and due to their rich sensing features, smartphones have become popular elements of modern human activity recognition systems. Besides activity recognition, smartphones have also been employed with success in fall detection/recognition systems, although a combined approach has not been evaluated yet. This article presents the results of a comprehensive evaluation of using a smartphone’s acceleration sensor for human activity and fall recognition, including 12 different types of activities of daily living (ADLs) and 4 different types of falls, recorded from 66 subjects in the context of creating “MobiAct”, a publicly available dataset for benchmarking and developing human activity and fall recognition systems. An optimized feature selection and classification scheme is proposed for each, a basic, i.e. recognition of 6 common ADLs only (99.9% accuracy), and a more complex human activity recognition task that includes all 12 ADLs and 4 falls (96.8% accuracy).