Human activity recognition and monitoring using smartphones

Abstract

The sedentary lifestyle is becoming popular especially for intellectual work. Although physical inactivity lifestyle may cause many unexpected illnesses, it is complicated to build up a positive lifestyle due to the lacks of reminder systems to manage and monitor physical activities of people. This research represents an effective way for daily activity monitoring using accelerator and gyroscope sensors embedded in a smartphone. Signals were recorded from accelerator and gyroscope sensors while a user wearing the smartphone performs different activities (going downstairs, going upstairs, sitting with the phone in a pocket, driving and putting the phone on the table). The classification algorithms with k-nearestneighbor (kNN) and artificial neural network (ANN) were applied to recognize user’s activities. The overall accuracy of recognizing five activities is 74% for kNN and 75.3% for ANN respectively. Based on the activities recognized during the day, users are able to manage their daily activities for a better life.