HARTH — Free-Living Human Activity Recognition Dataset
Catalog Summary
"6,461,328 annotated samples from 22 free-living participants wearing two accelerometers; CSVs contain eight core fields, with extra index columns in three files."
Catalog Notes
Overview
HARTH records free-living activity from 22 participants wearing lower-back and thigh accelerometers at 50 Hz for approximately two hours each. The source describes professional annotation using chest-mounted video.
Measured structure
The 22 CSVs contain 6,461,328 data rows. Eight core fields are timestamp, three back acceleration axes, three thigh acceleration axes and label. S015.csv and S021.csv additionally contain an index column; S023.csv has an unnamed leading index column. Thus 19 files have 8 columns and 3 have 9. Align columns by their meaning before concatenating files.
Interpretation and citation
Labels cover walking, running, shuffling, stairs, standing, sitting, lying and cycling variants. A single fixed column_count is omitted because the distributed schemas differ. The citation year 2023 denotes the verified UCI deposit date; UCI’s generated citation omits a year and its DOI metadata carries an invalid year value. It is not a claim that the underlying study was first published in 2023.
Source and provenance
Dataset DOI: https://doi.org/10.24432/C5NC90.
View Data Structure
To explore column names, data types, and sample rows, visit the official dataset page on UCI Machine Learning Repository.
Preview on UCI Machine Learning RepositoryCite This Dataset
Logacjov, A., Kongsvold, A., Bach, K., Bårdstu, H. B., & Mork, P. J. (2023). HARTH [Data set]. UCI Machine Learning Repository. https://doi.org/10.24432/C5NC90
Source metadata: UCI Machine Learning Repository (2023) · DOI: 10.24432/C5NC90
Indexed by IoTDataset.com on Sep 14, 2026
Review the Source Record
Confirm the licence, version, access conditions, file format, and provenance at the source before use.