Detection of the type of physical activity based on an IMU sensor
Paolino, Alessia
Promoteur(s) : Bruls, Olivier ; Schwartz, Cédric
Date de soutenance : 5-sep-2024/6-sep-2024 • URL permanente : http://hdl.handle.net/2268.2/20850
Détails
Titre : | Detection of the type of physical activity based on an IMU sensor |
Auteur : | Paolino, Alessia |
Date de soutenance : | 5-sep-2024/6-sep-2024 |
Promoteur(s) : | Bruls, Olivier
Schwartz, Cédric |
Membre(s) du jury : | Ruffoni, Davide
Drion, Guillaume |
Langue : | Anglais |
Nombre de pages : | 129 |
Mots-clés : | [en] Human Activity Recognition (HAR), [en] Wearable Sensors [en] Machine Learning Algorithms [en] Physical Activity Monitoring [en] Multilayer Perceptron (MLP) [en] Motion Analysis [en] Inertial Measurement Units (IMUs) |
Discipline(s) : | Ingénierie, informatique & technologie > Multidisciplinaire, généralités & autres |
Public cible : | Chercheurs Professionnels du domaine Etudiants |
Institution(s) : | Université de Liège, Liège, Belgique |
Diplôme : | Cours supplémentaires destinés aux étudiants d'échange (Erasmus, ...) |
Faculté : | Mémoires de la Faculté des Sciences appliquées |
Résumé
[en] The primary goal of this research is to develop a Human Activity Recognition (HAR) system using Inertial Measurement Units (IMUs), such as accelerometers and gyroscopes, to accurately identify and classify various types of physical movements.
The study specifically explores the use of wearable sensors for monitoring motor activities, offering an alternative solution to traditional camera-based motion capture systems, which have significant limitations, such as high costs and privacy concerns. The thesis discusses various stages of the process, including data acquisition through an experimental setup, data preprocessing, feature extraction and selection, and finally, the application of machine learning algorithms, such as Multilayer Perceptron (MLP) neural networks, for activity recognition and analysis.
The research also includes a comparative evaluation of the performance of models based on sensors positioned in different parts of the body (wrist, thigh, pocket) and provides detailed results regarding the accuracy of the models used.
Fichier(s)
Document(s)
Description: Thesis without the annexe, the section 6 is the Annexe A
Taille: 4.13 MB
Format: Adobe PDF
Description: -
Taille: 4.17 MB
Format: Adobe PDF
Annexe(s)
Citer ce mémoire
L'Université de Liège ne garantit pas la qualité scientifique de ces travaux d'étudiants ni l'exactitude de l'ensemble des informations qu'ils contiennent.