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 Concept Videos

You might also read

Related Articles

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

Sort by
Same author

Characterization and preliminary heterosis evaluation of novel wheat genetically divergent populations.

Frontiers in plant science·2026
Same author

Stratification of primary antiphospholipid syndrome by mechanistic immunophenotype: machine learning identifies distinct T-cell and T-bet+CD11c+ B cell-driven patient clusters.

Clinical and experimental rheumatology·2026
Same author

Mindful attention as a mechanism underlying non-suicidal self-injury: evidence from a longitudinal intervention trial.

BMC psychiatry·2026
Same author

A specialized population of hair afferents dedicated to transmitting mechanical itch.

Neuron·2026
Same author

Resting-state cortical activity, biomarkers and functional performance identify distinct biopsychosocial phenotypes in young adults with chronic postsurgical pain.

Frontiers in pain research (Lausanne, Switzerland)·2026
Same author

Nephrotoxicity of Evodiamine in Mice: Mechanistic Insights from Integrated Network Toxicology and Transcriptomic Profiling.

International journal of molecular sciences·2026

Related Experiment Video

Updated: Mar 15, 2026

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

5.6K

Lightweight Visual Dynamic Gesture Recognition System Based on CNN-LSTM-DSA.

Zhenxing Wang1, Ziyan Wu1, Ruidi Qi1

  • 1School of Computer and Information Engineering, Shanghai Polytechnic University, Shanghai 201209, China.

Sensors (Basel, Switzerland)
|March 14, 2026
PubMed
Summary

This study presents a lightweight visual dynamic gesture recognition system using a CNN-LSTM-DSA model for efficient deployment on embedded devices. The system achieves high accuracy for both static (96%) and dynamic (90.19%) gestures with low response delay.

Keywords:
CNN-LSTM hybrid modelbionic robotic handdepthwise separable convolutionvisual gesture recognition

More Related Videos

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

1.4K
Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines
08:27

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines

Published on: January 5, 2024

1.7K

Related Experiment Videos

Last Updated: Mar 15, 2026

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

5.6K
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

1.4K
Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines
08:27

Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines

Published on: January 5, 2024

1.7K

Area of Science:

  • Computer Vision
  • Robotics
  • Machine Learning

Background:

  • Large-scale gesture recognition models face challenges in computational complexity and embedded deployment.
  • Efficient and accurate gesture recognition is crucial for human-robot interaction and control systems.

Purpose of the Study:

  • To design and implement a lightweight visual dynamic gesture recognition system.
  • To address computational complexity and enable efficient deployment on embedded devices.
  • To achieve high-precision recognition of complex static and dynamic gestures.

Main Methods:

  • Utilized a lightweight Convolutional Neural Network (CNN)-Long Short-Term Memory (LSTM)-Dynamic Segment Assignment (DSA) model.
  • Employed MediaPipe for extracting 3D keypoint coordinates from camera-captured hand images.
  • Implemented joint angle calculation and sliding window smoothing for static gesture recognition.
  • Modeled keypoint time series using the CNN-LSTM-DSA hybrid model for dynamic gesture recognition.

Main Results:

  • Achieved static gesture recognition accuracy of up to 96% and dynamic gesture recognition accuracy of 90.19%.
  • Demonstrated robustness under various lighting and background conditions.
  • Maintained an overall response delay of less than 300 ms.
  • Ensured smooth servo motor angle mapping and stable robotic hand movements.

Conclusions:

  • The proposed lightweight CNN-LSTM-DSA system effectively overcomes challenges in gesture recognition model complexity and deployment.
  • The system offers a robust and accurate solution for both static and dynamic gesture recognition.
  • This approach facilitates efficient human-robot interaction and control applications on embedded systems.