Hand Gesture Recognition Using Compact CNN Via Surface Electromyography Signals.

Area of Science:

• Biomedical Engineering
• Machine Learning
• Signal Processing

Background:

• Surface Electromyography (sEMG) signals contain hidden features crucial for predicting human motion intention.
• Current deep neural network models for sEMG analysis often suffer from a large number of parameters, limiting their practical application.
• Efficient and accurate interpretation of sEMG signals is vital for advancements in prosthetics, human-computer interfaces, and rehabilitation robotics.

Purpose of the Study:

• To design a compact Convolution Neural Network (CNN) model for enhanced feature extraction from sEMG signals.
• To improve the classification accuracy of human motion intention prediction using sEMG data.
• To reduce the number of parameters in the sEMG analysis model without compromising performance.

Main Methods:

• Training a compact Convolution Neural Network (CNN) model on sEMG datasets.
• Utilizing the CNN for extracting hidden features from sEMG signals.
• Validating the model's performance on the Ninapro DB5 and Myo datasets for gesture recognition.

Main Results:

• The proposed compact CNN model demonstrated improved classification accuracy for gesture recognition.
• The model successfully reduced the number of parameters compared to existing deep learning approaches.
• Validation on Ninapro DB5 and Myo datasets confirmed the model's effectiveness in analyzing sEMG signals.

Conclusions:

• A compact CNN model offers an effective solution for extracting features from sEMG signals for motion intention prediction.
• The developed model achieves high classification accuracy with a reduced parameter count, making it suitable for real-world applications.
• This research contributes to more efficient and accurate sEMG-based human-computer interaction systems.