# GREFTUD Dataset

The *Gesture Recognition using EMG and FMG TU Darmstadt (GREFTUD)* dataset contains labeled recordings of electromyography (EMG) and forcemyography recordings of 13 healthy subjects, each of whom performed 66 distinct hand movements. 
The movements were labeled using a high speed camera and the start and end time (in ms) of each movement were noted manually. The dataset was presented in our paper "On the Benefit of FMG and EMG Sensor Fusion for Gesture Recognition Using Cross-Subject Validation". ([-> 10.1109/TNSRE.2025.3543649](https://doi.org/10.1109/TNSRE.2025.3543649))

The file ``dataset.hdf5`` contains the complete dataset, including the manual labels, but no video data.

An example on how to use the dataset can be found in ``hdf5_example.ipynb``.

## Label Naming Convention
The finger movements are labeled using the following naming convention : {Number of fingers moved}.{Movement ID}{f (Flexion) or e (Extension)}{s (slow) or n (normal) or q (quick)}. An example would be ``1.1fs`` which would correspond to a slow flexion of the thumb.

A table relating the labels/movement IDs to the fingers moved can be found in the ``dataset.hdf5`` file.

The wrist movements are labeled using the following naming convention : 6.{Movement ID}g{s (slow) or n (normal) or q (quick)}. The movement IDs for the wrist correspond to the following movements :

| Movement ID   | Movement |
| ------------- |:-------------:|
| 1      | Wrist extension |
| 2      | Wrist flexion     |
| 3 | Wrist abduction    |
| 4 | Wrist adduction      |

## Usage and Licensing

1. Setup a Python Environment using Miniconda
    ```
    conda create -n "greftud" python=3.10
    conda activate greftud
    pip install -r requirements.txt
    ```
    
2. Select the environment in a Jupyter-Notebook viewer of your choice


This work is licensed under a
[Creative Commons Attribution 4.0 International License][cc-by].

[![CC BY 4.0][cc-by-image]][cc-by]

[cc-by]: http://creativecommons.org/licenses/by/4.0/
[cc-by-image]: https://i.creativecommons.org/l/by/4.0/88x31.png
[cc-by-shield]: https://img.shields.io/badge/License-CC%20BY%204.0-lightgrey.svg

If you use this dataset in your research please cite our article:

```
@article{rohr.2025,
  author={Rohr, Maurice and Haidamous, Jad and Schäfer, Niklas and Schaumann, Stephan and Latsch, Bastian and Kupnik, Mario and Antink, Christoph Hoog},
  journal={IEEE Transactions on Neural Systems and Rehabilitation Engineering}, 
  title={On the Benefit of FMG and EMG Sensor Fusion for Gesture Recognition Using Cross-Subject Validation}, 
  year={2025},
  volume={33},
  number={},
  pages={935-944},
  keywords={Electromyography;Muscles;Gesture recognition;Force;Hands;Accuracy;Sensor fusion;Data acquisition;Cameras;Feature extraction;Electromyography (EMG);ferroelectrets;force myography (FMG);gesture recognition;sensor fusion},
  doi={10.1109/TNSRE.2025.3543649}}
```