Instance based human physical activity(hpa) recognition using shimmer2 wearable sensor data sets

Abstract

Human physical activity (HPA) recognition is one of the a large amount emerging fields of research in pervasive summing. In wearable computing scenarios, human physical activities such as standing still, sitting and relaxing, lying down, walking, climbing stairs, waist bends forward, front elevation of arms, knee bending, cycling, jogging, running and jump front and back can be implicit from sensor data provided by shimmer2 acceleration sensors. In such scenarios, most methods use a one or two dimensional features, nevertheless of which activity to be identified. This paper we can identified how to predict human physical activity using tri-accelerometer three dimensional data generated by shimmer2 wearable sensor device. We represent a efficient sensor data analysis of features computed from a realistic accelerometer sensor data set and different classifiers are studied on instances based data sets. This shows that the choice of time domain feature and the window dimension accomplished on which the computed features that transform the activity accuracy rates for different huamn physical activities.