Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jan 7, 2026

Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
08:15

Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision

Published on: March 28, 2025

1.1K

Hybrid Convolutional Vision Transformer for Robust Low-Channel sEMG Hand Gesture Recognition: A Comparative Study

Ruthber Rodriguez Serrezuela1, Roberto Sagaro Zamora2, Daily Milanes Hermosilla3

  • 1Department of Mechatronics Engineering, University Corporation of Huila, Neiva 410001, Colombia.

Biomimetics (Basel, Switzerland)
|December 24, 2025
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Robust Motor Imagery Tasks Classification Approach Using Bayesian Neural Network.

Sensors (Basel, Switzerland)·2023
Same author

Monte Carlo Dropout for Uncertainty Estimation and Motor Imagery Classification.

Sensors (Basel, Switzerland)·2021
See all related articles

A Convolutional Vision Transformer (CViT) shows superior performance over a classical Convolutional Neural Network (CNN) for hand gesture classification using low-channel surface electromyography (sEMG) signals. This advancement is key for developing compact prosthetic control systems.

Area of Science:

  • Biomedical Engineering
  • Machine Learning
  • Rehabilitation Technology

Background:

  • Surface electromyography (sEMG) is crucial for prosthetic control and human-machine interaction.
  • Existing sEMG studies often use high-density recordings or extensive gesture sets, limiting data on low-channel, reduced-gesture systems.
  • Compact sEMG systems require efficient classification models for practical prosthetic applications.

Purpose of the Study:

  • To benchmark a classical Convolutional Neural Network (CNN) against a Convolutional Vision Transformer (CViT) for hand gesture classification in low-channel sEMG systems.
  • To evaluate model performance on both non-amputee and transradial amputee datasets.
  • To determine the robustness and generalization capabilities of CNN and CViT architectures under challenging sEMG conditions.

Main Methods:

Keywords:
Vision Transformer (ViT)convolutional neural network (CNN)hand gesture recognitionhybrid deep learningmyoelectric pattern recognitionsurface electromyography (sEMG)

Related Experiment Videos

Last Updated: Jan 7, 2026

Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
08:15

Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision

Published on: March 28, 2025

1.1K
  • Two datasets were used: a proprietary 8-channel Myo dataset and a 12-channel NinaPro DB3 subset from transradial amputees.
  • Both CNN and CViT models were trained using identical standardized preprocessing, segmentation, and balanced windowing.
  • Model accuracy was compared across datasets, focusing on performance in low-channel, heterogeneous sEMG signal scenarios.

Main Results:

  • The CNN achieved 94.2% accuracy on the Myo dataset but showed variability (92.0% accuracy) on the amputee NinaPro dataset.
  • The CViT model consistently matched or exceeded CNN performance, achieving 96.6% accuracy on Myo and 94.2% on NinaPro.
  • Statistical analysis indicated significant performance differences on the Myo dataset, with CViT demonstrating superior results.

Conclusions:

  • The Convolutional Vision Transformer (CViT) offers enhanced robustness and generalization compared to classical CNNs for low-channel sEMG hand gesture classification.
  • CViT's performance is particularly advantageous in heterogeneous signal conditions, such as those encountered with amputee recordings.
  • These findings highlight the suitability of CViT for developing advanced, compact prosthetic control systems.