Human activity classification with miniature inertial and magnetic sensor signals

This study provides a comparative performance assessment of various pattern recognition techniques on classifying human activities that are performed while wearing miniature inertial and magnetic sensors. Activities are classified using five sensor units worn on the chest, the arms, and the legs. Each sensor unit comprises a tri-axial accelerometer, a tri-axial gyroscope, and a tri-axial magnetometer. The classification techniques compared in this study are: naive Bayesian (NB), artificial neural networks (ANN), dissimilarity-based classifier (DBC), various decision-tree algorithms, Gaussian mixture model (GMM), and support vector machines (SVM). The methods that result in the highest correct differentiation rates are found to be GMM (99.1%), ANN (99.0%), and SVM (98.9%).