Related Experiment Video
Updated: Jan 16, 2026
08:15
Capturing Dynamic Finger Gesturing with High-resolution Surface Electromyography and Computer Vision
Published on: March 28, 2025
1.2K
Finger joint angle and gesture estimation under natural conditions with a soft printed electrode array
Nitzan Luxembourg1, Rufael Fekadu Marew2, Dvir Teitelbaum1
1School of Electrical and Computer Engineering, Tel Aviv University, Tel Aviv, Israel.
APL Bioengineering
|October 6, 2025
Summary
Surface electromyography (sEMG) enables finger gesture recognition for human-machine interfaces. This study enhances sEMG accuracy for both static and dynamic hand gestures using a novel sensor and AI model.
Area of Science:
- Biomedical Engineering
- Human-Computer Interaction
- Wearable Technology
Background:
- Surface electromyography (sEMG) is a promising technology for human-machine interfaces, particularly when visual methods are impractical.
- sEMG-based gesture recognition faces challenges due to movement artifacts, individual muscle variations, and hand position changes, limiting dynamic gesture recognition.
- Existing research predominantly focuses on static hand positions, hindering real-world applications.
Purpose of the Study:
- To develop and evaluate an integrated system for finger gesture recognition using soft wearable sEMG sensors.
- To improve the accuracy and robustness of sEMG-based gesture recognition across both static and dynamic hand positions.
- To assess the potential of a Video-Vision-Transform model combined with motion sensor training for predicting finger joint angles and recognizing gestures.
Main Methods:
- Integration of a soft wearable sEMG sensor with a Video-Vision-Transform (ViT) model.
- Utilizing motion sensor-based training to enhance gesture recognition capabilities.
- Testing the system's performance in differentiating finger angles and recognizing gestures in static and dynamic conditions.
Main Results:
- The integrated system demonstrated the ability to differentiate finger angles and recognize gestures across subjects.
- Recognition accuracy reached 0.85 for static and 0.87 for dynamic gestures in the best-performing subject.
- The system showed stable performance across both static and dynamic conditions, despite inter-subject variability.
Conclusions:
- This study advances sEMG-based finger gesture recognition by achieving stable performance in both static and dynamic scenarios.
- The developed approach shows potential for natural and real-world human-machine interface applications.
- Further research may address inter-subject variability to broaden applicability.
