Surface Electromyography (sEMG) Dataset for Pectoralis Muscle

The dataset consists of 29 sEMG signals recorded from the pectoralis muscle of 18 healthy adults with varied age, weight, and height, after obtaining informed consent, at the LINS Laboratory (USTHB).

The main objective is to identify two differente classes, rest  and muscle contraction.

Recordings were made using gel electrodes and a custom EMG acquisition board at a sampling rate of 1000 Hz, transmitted in real time to a computer via ESP32 and MATLAB, and stored in CSV format. Each session lasted 65 seconds and followed a structured protocol where participants alternated between the two classes every few seconds. Most participants contributed two sessions with a 3.7 V supply, 

Each CSV file contains four columns: 

Sample

Time 

FilteredV 

Label

The custom EMG acquisition board was designed to provide clean, high-quality signals through the following stages:

Instrumentation amplifier with a gain of 500.

Active band-pass filter (30–500 Hz) to suppress motion artifacts, DC offset, and high-frequency noise.

Notch filter at 50 Hz to remove power line interference.

Main amplification stage with a gain of 201.

Digital first-order IIR high-pass filter (cutoff 50 Hz)  to mitigate ECG contamination.

Dataset Summary

Type of Record
Number of Subjects
Number of Signals
Sampling Rate
Duration per Signal

EMG signals
18
29
1000 Hz
65 s

Note:

The dataset contains well-labeled signals with two classes: Class 0 = rest and Class 1 = contraction.

It is intended for use in the development of bionic control systems.

A large CSV file containing all signals concatenated has been added for convenience.

An additional CSV file without labels is included for testing purposes after training.

For more details, please feel free to contact us:

[e-mail]

[e-mail]

[e-mail]

[e-mail]