license: apache-2.0
language:
- en
tags:
- timeseries
- wearables
pretty_name: Smartwatch Data For Wearable Motion Capture
size_categories:
- 100K<n<1M
WearMoCap Dataset
TLDR: This dataset comprises optical motion capture data of left arm motions and body orientations, which are paired with smartwatch and phone sensor measurements. All synced, calibrated, and preprocessed time series data is stored as CSVs.
We collected this data to develop and test the open-source library WearMoCap, a system facilitating multimodal pose tracking from smartwatch and smartphone data. Its source code and tutorials are available on github. Details about data composition and collection setups are described in a published technical report here. A video showcasing the WearMoCap system is available here.
Repository Structure
As described in the technical report, we partitioned data for operating WearMoCap in three modes: Watch Only, Phone Uarm, and Phone Pocket. Each mode is one subfolder, which contains a README with more information and subfolders spliting the data by human subjects.
Simplified directory structure overview:
mode
|-- README.md
|-- subj01
|-- data01.csv
|-- [..]
|-- subj02
|-- [..]
[..]