There were 21 able-bodied participants in this study. Approval of the experiment was granted by the Research Ethics Committee of Xidian University.
A Myo armband, equipped with sEMG and IMU sensors, was used to capture gestures activity. After wearing the Myo armband, each participant was required to perform a hardware calibration procedure. The procedure is to straighten the five fingers together and bend the wrist outward by 90 degrees until the MYO armband cease vibrating.
The eight-channel sEMG signals were collected at a 200-Hz sampling rate. The three-channel acceleration signals (Acc) and three-channel gyroscope signals (Gyro) were captured at a 50-Hz sampling rate.
During the data acquisition, each type of gesture was executed in six repetitive trials. To prevent muscle fatigue, there was a 5 s rest between trials.
Participants performed each static gesture within 5 s and each dynamic gesture within 3 s.
| Folder | Label | Class | Name | Duration(s) |
|---|---|---|---|---|
| A | 1 | static | Neutral | 5 |
| A | 2 | static | Redial Deviation | 5 |
| A | 3 | static | Flexion | 5 |
| A | 4 | static | Ulnar Deviation | 5 |
| A | 5 | static | Extension | 5 |
| A | 6 | static | Hand Close | 5 |
| A | 7 | static | Hand Open | 5 |
| A | 8 | dynamic | Yaw Right | 3 |
| A | 9 | dynamic | Yaw Left | 3 |
| A | 10 | dynamic | Pitch Above | 3 |
| A | 11 | dynamic | Pitch Below | 3 |
| A | 12 | dynamic | Roll Exterior | 3 |
| A | 13 | dynamic | Roll Interior | 3 |
| B | 1 | static | OK | 5 |
| B | 2 | static | Yeah | 5 |
| B | 3 | static | Six | 5 |
| B | 4 | static | Nice | 5 |
| B | 5 | dynamic | NO | 5 |
| B | 6 | dynamic | Come | 5 |
| B | 7 | dynamic | Pinch | 5 |
| B | 8 | dynamic | Raise Hand | 2 |
| B | 9 | dynamic | Release Hand | 2 |
| B | 10 | dynamic | Wave Hand | 2 |
0 indicates a rest phase between the same gestures, and -1 indicates a rest phase between different gestures.
The Paper had been published in IEEE Sensors Journal, and utilized Dataset A.
https://doi.org/10.1109/JSEN.2023.3327999
In XDMyo-A, the raw signal data of the 21 subjects included 1,638 (13 × 21 × 6) gestures. The dataset was partitioned into the source-dataset and the target-dataset. The source-dataset comprised data from 15 participants.
G. Li, B. Wan, K. Su, J. Huo, C. Jiang and F. Wang, "sEMG and IMU Data-Based Hand Gesture Recognition Method Using Multistream CNN With a Fine-Tuning Transfer Framework," in IEEE Sensors Journal, vol. 23, no. 24, pp. 31414-31424, 15 Dec.15, 2023, doi: 10.1109/JSEN.2023.3327999.
@ARTICLE{10305519,
author={Li, Guiyin and Wan, Bo and Su, Kejia and Huo, Jiwang and Jiang, Changhua and Wang, Fei},
journal={IEEE Sensors Journal},
title={sEMG and IMU Data-Based Hand Gesture Recognition Method Using Multistream CNN With a Fine-Tuning Transfer Framework},
year={2023},
volume={23},
number={24},
pages={31414-31424},
doi={10.1109/JSEN.2023.3327999}}Guiyin Li, Bo Wan, Kejia Su, Jiwang Huo, Changhua Jiang, Fei Wang