Loading...

Hand Gesture Recognition using sEMG in Man-Machine Interfaces

Ameri Haftador, Monireh | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58007 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Jahed, Mehran
  7. Abstract:
  8. Fast and accurate recognition of hand gestures remains a significant challenge in the field of human-machine interfaces (HMI). This challenge is particularly critical in applications such as rehabilitation exoskeletons, which demand high accuracy. Surface electromyography (sEMG) signals are effective tools for recognizing hand and wrist movements due to their non-invasive acquisition and direct correlation with muscle activity. However, achieving both high accuracy and speed under diverse conditions continues to be a persistent challenge. In this study, classical machine learning and deep learning models—including KNN, Naïve Bayes, Decision Tree, Random Forest, SVM, LDA, MLP, and CNN—were evaluated for hand and wrist gesture recognition. Three datasets were utilized: one general dataset with four classes, and two specialized datasets comprising simple movements (thumb and index finger touch) with two classes. The data were preprocessed into three formats—processed signals, extracted features, and short-term Fourier transform—and then fed into the classifiers. The results indicated that all classifiers could perform classification in under 200 milliseconds, making them suitable for HMI applications. The best performance for raw data was achieved using a support vector machine (SVM) with extracted features, attaining an accuracy of 84%. For the EMG-EPN-612 dataset, the CNN model using short-term Fourier transform provided the best results, with accuracies of 92% respectively. These findings highlight the critical importance of selecting suitable methods based on application requirements and system constraints, particularly for real-time systems with limited processing capabilities
  9. Keywords:
  10. Surface Electromyogram ; Neural Network ; Machine Learning ; Hand Gestures ; Hand And Wrist Gesture Recognition ; Human-Machine Interaction

 Digital Object List

 Bookmark

No TOC