MyoUP

MyoUP Dataset Overview

Dataset Overview

• Purpose: Developed the MyoUP database to obtain sEMG data without professional calibration.
• Inspiration: Inspired by the Ninapro database; many recorded gestures correspond to those in Ninapro.
• Equipment: Data were collected using the Myo Armband (Thalmic Labs), which provides 200 Hz sampling and eight dry sEMG channels.

Dataset Information

• Participants: Eight complete subjects (3 female, 5 male; 1 left‑handed, 7 right‑handed; mean age 22.38 ± 1.06 years).
• Data Acquisition: Divided into three parts, covering five basic finger motions, twelve isometric/isotonic hand configurations, and five grasp gestures.
• Acquisition Procedure: Each gesture was repeated five times, each repetition lasting 5 seconds with a 5‑second inter‑trial interval to prevent muscle fatigue.

Dataset Applications

• Gesture Recognition: Real‑time gesture‑recognition models have been developed based on the MyoUP dataset using convolutional neural networks (CNNs).

Citation

• Tsagkas, N., Tsinganos, P., & Skodras, A. (2019). "On the Use of Deeper CNNs in Hand Gesture Recognition Based on sEMG Signals." 2019 10th International Conference on Information, Intelligence, Systems and Applications (IISA), pp. 1‑4. doi: 10.1109/IISA.2019.8900709.