A Triaxial Accelerometer-Based Physical-Activity Recognition via Augmented-Signal Features and a Hierarchical Recognizer

Abstract

Physical-activity recognition via wearable sensors can provide valuable information regarding an individual's degree of functional ability and lifestyle. In this paper, we present an accelerometer sensor-based approach for human-activity recognition. Our proposed recognition method uses a hierarchical scheme. At the lower level, the state to which an activity belongs, i.e., static, transition, or dynamic, is recognized by means of statistical signal features and artificial-neural nets (ANNs). The upper level recognition uses the autoregressive (AR) modeling of the acceleration signals, thus, incorporating the derived AR-coefficients along with the signal-magnitude area and tilt angle to form an augmented-feature vector. The resulting feature vector is further processed by the linear-discriminant analysis and ANNs to recognize a particular human activity. Our proposed activity-recognition method recognizes three states and 15 activities with an average accuracy of 97.9% using only a single triaxial accelerometer attached to the subject's chest.

Action Graph - A Versatile Data Structure for Action Recognition

“…We refer to (Latré et al, 2011) for a recent survey of wireless body area networks and to (Khan et al, 2010) for a recent accelerometer based physical activity system. Although not all of these are suitable for our approach, many of them present promising perspectives.…”

“…Within this context, they analyze the quality of known classifiers for recognizing activities with particular emphasis on the importance of selected features and level of difficulty of recognizing specific activities. The system developed by (Khan et al, 2010) is capable of recognizing a broad set of human physical activities using only a single triaxial accelerometer. The approach is of higher accuracy than the previous works due to a novel augmented-feature vector.…”

Activity and Emotion Recognition for Elderly Health Monitoring

“…The physical activity recognition in [10,24] utilized a single triaxial accelerometer to distinguish between the different ADL of relatively old persons (six males, two females, age: mean = 65, SD = 3 years old). The a triaxial accelerometer was attached to 5 different places of the body, the position of chest, a particular orientation, was found to be very practical and was able to classify fifteen activities with an average accuracy of 97.9%.…”

“…They can be divided into three categories related on the type of data they are based on: i) Motion-sensor-based methods [7,[10][11][12][13], attached or wearable, they utilize on body sensors like the accelerometer and gyroscope, to sense the movements of body parts. ii) Radio-based methods [14,15], the wireless radio types include: ZigBee; which build a small network of sensors on the body.…”

Posture transition identification on PD patients through a SVM-based technique and a single waist-worn accelerometer

“…First, Artifical Neural Networks (ANN) have been used [34]. ANN establishes a mathematical relation between the outputs and the inputs and one of the most used methods in activity and posture recognition is the Multilayer Layer Perceptron [35], which is the employed method in the present work and it is trained by means of the backpropagation algorithm with three different learning rates.…”