Human activity recognition based on machine learning classification of smartwatch accelerometer dataset
This paper presents two Machine Learning models that classify time series data given from smartwatch accelerometer of observed subjects. For the purpose of classification we use Deep Neural Network and Random Forest classifier algorithms. The comparison of both models shows that they have similar pe...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
University of Belgrade - Faculty of Mechanical Engineering, Belgrade
2021-01-01
|
Series: | FME Transactions |
Subjects: | |
Online Access: | https://scindeks-clanci.ceon.rs/data/pdf/1451-2092/2021/1451-20922101225R.pdf |
id |
doaj-042bc27ebef74825adba420854eecb94 |
---|---|
record_format |
Article |
spelling |
doaj-042bc27ebef74825adba420854eecb942021-01-24T11:13:18ZengUniversity of Belgrade - Faculty of Mechanical Engineering, BelgradeFME Transactions1451-20922406-128X2021-01-014912252321451-20922101225RHuman activity recognition based on machine learning classification of smartwatch accelerometer datasetRadivojević Dušan S.0Mirkov Nikola S.1https://orcid.org/0000-0001-6826-2857Maletić Slobodan2University of Belgrade, "VINČA" Institute of Nuclear Sciences - National Institute of the Republic of Serbia, Department of Thermal engineering and Energy, Belgrade, SerbiaUniversity of Belgrade, "VINČA" Institute of Nuclear Sciences - National Institute of the Republic of Serbia, Department of Thermal engineering and Energy, Belgrade, SerbiaUniversity of Belgrade, "VINČA" Institute of Nuclear Sciences - National Institute of the Republic of Serbia, Department of Thermal engineering and Energy, Belgrade, SerbiaThis paper presents two Machine Learning models that classify time series data given from smartwatch accelerometer of observed subjects. For the purpose of classification we use Deep Neural Network and Random Forest classifier algorithms. The comparison of both models shows that they have similar performance with regard to recognition of subject's activities that are used in the test group of the dataset. Training accuracy reaches approximately 95% and 100% for Deep Learning and Random Forest model respectively. Since the validation and recognition, reached about 81% and 75% respectively, a tendency for improving accuracy as a function of number of participants is considered. The influence of data sample precision to the accuracy of the models is examined since the input data could be given from various wearable devices.https://scindeks-clanci.ceon.rs/data/pdf/1451-2092/2021/1451-20922101225R.pdfhuman activity recognitionmachine learningsmartwatch |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Radivojević Dušan S. Mirkov Nikola S. Maletić Slobodan |
spellingShingle |
Radivojević Dušan S. Mirkov Nikola S. Maletić Slobodan Human activity recognition based on machine learning classification of smartwatch accelerometer dataset FME Transactions human activity recognition machine learning smartwatch |
author_facet |
Radivojević Dušan S. Mirkov Nikola S. Maletić Slobodan |
author_sort |
Radivojević Dušan S. |
title |
Human activity recognition based on machine learning classification of smartwatch accelerometer dataset |
title_short |
Human activity recognition based on machine learning classification of smartwatch accelerometer dataset |
title_full |
Human activity recognition based on machine learning classification of smartwatch accelerometer dataset |
title_fullStr |
Human activity recognition based on machine learning classification of smartwatch accelerometer dataset |
title_full_unstemmed |
Human activity recognition based on machine learning classification of smartwatch accelerometer dataset |
title_sort |
human activity recognition based on machine learning classification of smartwatch accelerometer dataset |
publisher |
University of Belgrade - Faculty of Mechanical Engineering, Belgrade |
series |
FME Transactions |
issn |
1451-2092 2406-128X |
publishDate |
2021-01-01 |
description |
This paper presents two Machine Learning models that classify time series data given from smartwatch accelerometer of observed subjects. For the purpose of classification we use Deep Neural Network and Random Forest classifier algorithms. The comparison of both models shows that they have similar performance with regard to recognition of subject's activities that are used in the test group of the dataset. Training accuracy reaches approximately 95% and 100% for Deep Learning and Random Forest model respectively. Since the validation and recognition, reached about 81% and 75% respectively, a tendency for improving accuracy as a function of number of participants is considered. The influence of data sample precision to the accuracy of the models is examined since the input data could be given from various wearable devices. |
topic |
human activity recognition machine learning smartwatch |
url |
https://scindeks-clanci.ceon.rs/data/pdf/1451-2092/2021/1451-20922101225R.pdf |
work_keys_str_mv |
AT radivojevicdusans humanactivityrecognitionbasedonmachinelearningclassificationofsmartwatchaccelerometerdataset AT mirkovnikolas humanactivityrecognitionbasedonmachinelearningclassificationofsmartwatchaccelerometerdataset AT maleticslobodan humanactivityrecognitionbasedonmachinelearningclassificationofsmartwatchaccelerometerdataset |
_version_ |
1724326498549104640 |