Related Experiment Video
Updated: Jun 23, 2026
08:15
Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
Published on: March 28, 2025
EMG-based hand gesture recognition using multi-scale deep residual network with SE-module
Ramkumar Sivasakthivel1, Rajendran Thavasimuthu2,3, Manikandan Rajagopal4
1Department of Computer Science, School of Sciences, Christ University, Bangalore, Karnataka, India.
Scientific Reports
|June 20, 2026
Summary
This study introduces a new deep learning model for hand gesture recognition using EMG signals. The Multi-Scale Deep Residual Network achieves over 99% accuracy, significantly improving human-machine interaction.
Area of Science:
- Biomedical Engineering
- Machine Learning
- Human-Computer Interaction
Background:
- Hand Gesture Recognition (HGR) using Electromyography (EMG) signals is crucial for advancing human-machine interaction.
- Existing deep learning methods for HGR face challenges in modeling multi-scale temporal variations and channel-wise features in EMG data.
- Effectively capturing both temporal dependencies and feature relevance is key to improving HGR performance.
Purpose of the Study:
- To propose a novel deep learning-based Multi-Scale Deep Residual Network (DRN) integrated with a Squeeze-and-Excitation (SE) model for enhanced hand gesture recognition using EMG signals.
- To address the limitations of current methods in jointly modeling multi-scale temporal dynamics and channel-wise feature importance.
- To evaluate the proposed model's effectiveness on the EMG-EPN-612 dataset.
Main Methods:
- EMG signal data collection and preprocessing, including moving average filtering, min-max normalization, and sliding-window segmentation.
- Development of a Multi-Scale Deep Residual Network (DRN) incorporating SE-based channel recalibration for capturing multi-scale temporal patterns and feature importance.
- Training and testing the model on the EMG-EPN-612 dataset, with 75% for training and 25% for testing.
Main Results:
- The proposed Multi-Scale DRN with SE module achieved high performance metrics: 99.24% accuracy, 99.15% precision, 99.17% F1-score, 99.10% specificity, and 99.20% recall.
- The model demonstrated superior performance compared to existing methods.
- The architecture effectively captured discriminative multi-scale temporal features and adaptively emphasized informative EMG channels.
Conclusions:
- The developed DL-based Multi-Scale DRN with SE model offers a significant advancement in EMG-based hand gesture recognition.
- The model's ability to jointly learn temporal patterns at different scales and channel importance leads to superior performance.
- This approach holds promise for more sophisticated and intuitive human-machine interaction systems.
