Related Experiment Video
Updated: Oct 21, 2025
08:15
Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
Published on: March 28, 2025
854
A Hierarchical View Pooling Network for Multichannel Surface Electromyography-Based Gesture Recognition
Wentao Wei1, Hong Hong2, Xiaoli Wu1
1School of Design Arts and Media, Nanjing University of Science and Technology, Nanjing, Jiangsu, China.
Computational Intelligence and Neuroscience
|September 6, 2021
Summary
This study introduces a novel hierarchical view pooling network (HVPN) for improved surface electromyography (sEMG) based hand gesture recognition. The HVPN framework enhances accuracy in both intrasubject and intersubject recognition tasks.
Area of Science:
- Biomedical Engineering
- Rehabilitation Engineering
- Machine Learning
Background:
- Surface electromyography (sEMG) is crucial for hand gesture recognition in biomedical and rehabilitation engineering.
- High-density sEMG (HD-sEMG) has advanced gesture recognition, but robust recognition with sparse multichannel sEMG remains challenging.
Purpose of the Study:
- To present a hierarchical view pooling network (HVPN) framework for enhanced multichannel sEMG-based gesture recognition.
- To improve gesture recognition by learning both view-specific and view-shared deep features from multiview feature spaces.
Main Methods:
- Developed a hierarchical view pooling network (HVPN) within a multiview deep learning context.
- Evaluated the HVPN framework on the NinaPro database using intrasubject and intersubject evaluations with 200 ms sliding windows.
Main Results:
- The proposed HVPN framework achieved high intrasubject accuracies (up to 90.3%) and intersubject accuracies (up to 88.9%) across five NinaPro subdatabases.
- HVPN significantly outperformed state-of-the-art methods in multichannel sEMG-based gesture recognition.
Conclusions:
- The HVPN framework offers a robust and effective solution for multichannel sEMG-based hand gesture recognition.
- This approach advances the capabilities of non-invasive prosthetics and human-computer interaction systems.
