6-subject (5M, 1F, 23 - 31 years old) dataset for continuous estimation of 2 DOFs: hand open/close (DOF 1) and wrist rotation (DOF 2). Each participant provided 3 distinct sets: 2 emulated behaviours, namely "All or Nothing" and "Anticipation", and a baseline set of data. Check out our MEC publication for an in-depth look at data collection details: https://conferences.lib.unb.ca/index.php/mec/article/view/2507.
We recommend that this dataset is used with our open-source toolkit, LibEMG, publicly available on GitHub. This dataset can be imported, downloaded, and ready for use in a few lines of code when used with LibEMG.
from libemg.datasets import UserComplianceDataset
odh = UserComplianceDataset().prepare_data()The data structure follows the format behaviour/subject/movement, where explanations of each component are provided below:
behaviour: Type of behaviour (e.g.,all-or-nothing,anticipation, orbaseline).subject: Subject ID (e.g.,subject-001).movement: Active DOF during these repetitions (e.g.,open-closefor hand open/close andpro-supfor wrist pronation/supination).
A labels.txt file is provided in each directory representing the ground truth shown to the participant. Each row represents a frame in an animation and each column represents a DOF - column 1 is hand open (+)/close (-) and column 2 is wrist pronation (-)/supination (+).
Data were collected using the HD-EMG EMaGer cuff from Université Laval. Device details are shown below:
- Name: EMaGer
- Sampling frequency: 1010 Hz
- Channels: 64
If you are using this dataset in any publications, please cite https://conferences.lib.unb.ca/index.php/mec/article/view/2507 for the dataset and https://ieeexplore.ieee.org/document/10340612 for use of the EMaGer cuff.