Year: 2013
Type: dissertation
Abstract: The demand for objectivity in clinical diagnosis has been one of the greatest challenges in Biomedical Engineering.The study, development and implementation of solutions that may serve as ground truth in Physical Activity (PA) recognition and in medical diagnosis of chronic motor diseases is ever more imperative.This thesis describes a Human Activity Recognition (HAR) framework based on feature extraction and feature selection techniques where a set of time, statistical and frequency domain features taken from 3dimensional accelerometer sensors was designed.In the present study, machine learning algorithms were applied in a non-supervised environment using feature representation of accelerometer data to discover the activities performed by different subjects.A feature selection framework is developed in order to improve the clustering accuracy and reduce computational costs.The features which best distinguish a particular set of activities were selected from a 180 th -dimensional feature vector through machine learning algorithms.The implemented framework achieved very encouraging results in human activity recognition: an average person-dependent Adjusted Rand Index (ARI) of 99.29% ± 0.5% and a person-independent ARI of 88.57% ± 4.0% were reached.Accurate and detailed measurement of an individual's PA is a key requirement for helping researchers understand the relationship between motor activity and health.
Author: Inês Machado
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: Fundação Champalimaud
OpenAlex ID: W2243626542