HandWave: An EMG-Powered System for Intuitive Gesture Recognition

Hand gestures are non-verbal communication techniques that use specific movements of hand, finger, and wrist postures to express ideas, feelings, or instructions. The classification of hand gestures accurately is challenging due to the unpredictability and large dimensionality of the data. The main objective of this study is to develop a classification model based on the EMG signals, which will identify the most significant features in order to optimize the classification accuracy. The proposed framework utilizes the MYO thalmic bracelet recording dataset. The process comprises two stages: the initial step includes extracting features, while the second stage proposes a grasshopper (GH)-based feature selection algorithm that integrates swarm intelligence (SI) and machine learning (ML) for the optimal selection of features. Thereafter, the best set of features is trained and categorized for hand gestures using classifiers like Decision Trees (DT), K-nearest neighbor (KNN), Artificial Neural Networks (ANN), Naive Bayes (NB), and ensemble bagged trees to see how well it does in terms of F-measure, precision, recall and accuracy. Overall, simulation results revealed that the proposed work based on the GH optimization algorithm, provides 99.98% accuracy using the KNN classifier and outperformed other classifiers in predicting hand gestures. This study has successfully proposed a robust EMG-based hand gesture classification model with high accuracy using the GH feature selection method. These results emphasized the capability of the KNN classifier in recognizing hand gestures and further extended valuable insights into the techniques of non-verbal communication and their applications in medical diagnostics and biomechanics.