HARTH
Donated on 2/21/2023
The Human Activity Recognition Trondheim (HARTH) dataset is a professionally-annotated dataset containing 22 subjects wearing two 3-axial accelerometers for around 2 hours in a free-living setting. The sensors were attached to the right thigh and lower back. The professional recordings and annotations provide a promising benchmark dataset for researchers to develop innovative machine learning approaches for precise HAR in free living.
Dataset Characteristics
Multivariate, Time-Series
Subject Area
Computer Science
Associated Tasks
Classification
Attribute Type
Real
# Instances
6461328
# Attributes
8
Information
For what purpose was the dataset created?
The dataset was created to train machine learning classifiers for human activity recognition based on professional annotations of activities in a free-living setting.
Who funded the creation of the dataset?
NTNU Helse
Additional Information
The HARTH dataset contains recordings of 22 participants wearing two 3-axial Axivity AX3 accelerometers for around 2 hours in a free-living setting. One sensor was attached to the right front thigh and the other to the lower back. The provided sampling rate is 50Hz. Video recordings of a chest-mounted camera were used to annotate the performed activities frame-by-frame. Each subject's recordings are provided in a separate .csv file. One such .csv file contains the following columns: 1. timestamp: date and time of recorded sample 2. back_x: acceleration of back sensor in x-direction (down) in the unit g 3. back_y: acceleration of back sensor in y-direction (left) in the unit g 4. back_z: acceleration of back sensor in z-direction (forward) in the unit g 5. thigh_x: acceleration of thigh sensor in x-direction (down) in the unit g 6. thigh_y: acceleration of thigh sensor in y-direction (right) in the unit g 7. thigh_z: acceleration of thigh sensor in z-direction (backward) in the unit g 8. label: annotated activity code The dataset contains the following annotated activities with the corresponding coding: 1: walking 2: running 3: shuffling 4: stairs (ascending) 5: stairs (descending) 6: standing 7: sitting 8: lying 13: cycling (sit) 14: cycling (stand) 130: cycling (sit, inactive) 140: cycling (stand, inactive)
Citation Requests/Acknowledgements
[1] A. Logacjov, K. Bach, A. Kongsvold, H. B. Bårdstu, and P. J. Mork, “HARTH: A Human Activity Recognition Dataset for Machine Learning,” Sensors, vol. 21, no. 23, Art. no. 23, Jan. 2021, doi: 10.3390/s21237853. [2] K. Bach et al., “A Machine Learning Classifier for Detection of Physical Activity Types and Postures During Free-Living,” Journal for the Measurement of Physical Behaviour, vol. 1, no. aop, pp. 1–8, Dec. 2021, doi: 10.1123/jmpb.2021-0015.
Introduction Paper
-
Logacjov,Aleksej, Logacjov,Aleksej, Kongsvold,Atle, Bach,Kerstin, Bårdstu,Hilde Bremseth & Mork,Paul Jarle. (2023). HARTH. UCI Machine Learning Repository.
@misc{misc_harth_779,
author = {Logacjov,Aleksej, Logacjov,Aleksej, Kongsvold,Atle, Bach,Kerstin, Bårdstu,Hilde Bremseth & Mork,Paul Jarle},
title = {{HARTH}},
year = {2023},
howpublished = {UCI Machine Learning Repository}
}Keywords
Creators
Aleksej Logacjov
aleksej.logacjov@ntnu.no
Norwegian University of Science and Technology
Aleksej Logacjov
aleksej.logacjov@ntnu.no
Norwegian University of Science and Technology
Atle Kongsvold
Norwegian University of Science and Technology
Kerstin Bach
Norwegian University of Science and Technology
Hilde Bremseth Bårdstu
Norwegian University of Science and Technology
Paul Jarle Mork
Norwegian University of Science and Technology
License
This dataset is licensed under a Creative Commons Attribution 4.0 International (CC BY 4.0) license.
This allows for the sharing and adaptation of the datasets for any purpose, provided that the appropriate credit is given.