Efficient Model for Enhancing Hand Gesture Identification Using Machine Learning with Principal Component Analysis (PCA) Feature Selection Technique

Abstract

Hand gesture recognition and classification play a vital role in human-computer interaction, offering intuitive and natural means of communication and control in various applications. This research investigates an innovative approach to build a model for the hand gesture image classification and enhance the accuracy and efficiency of hand gesture recognition systems using a machine learning-based method coupled with Principal Component Analysis (PCA). The proposed methodology aims to address the challenges associated with variability in hand gestures, environmental conditions, and computational complexity. The research leverages different machine learning techniques like Stochastic Gradient Descent (SGD), K-Nearest Neighbor (KNN), Decision Tree (DT), Random Forest (RF) and Gaussian Naïve Bayes (GNB) algorithm sand compares the classification performance with each other. Additionally, Principal Component Analysis (PCA) is employed for dimensionality reduction, extracting the most salient features from high-dimensional hand gesture data while preserving crucial information. The proposed model is trained and evaluated on the dataset, comparing its performance with different machine learning based classifiers which are used with the PCA feature selection technique. Furthermore, the impact of PCA-based feature selection on classification accuracy and computational efficiency is thoroughly analyzed. Using the Leap gesture dataset, the suggested method achieves remarkable results with an accuracy of 97.48% by KNN and also RFT and GNB achieve 99.99% and 82.09%, respectively. According to the experimental findings, the recommended PCA-based feature selection technique beat other classifiers in terms of F-measure, accuracy, recall, and precision.