An Efficient Sparse Gabor Descriptor Framework for Sign Language Recognition

Recognising person sign language for videos through algorithms remains a significant difficulty in computer vision (CV). The challenge of recognising human hand gestures with computer vision might be solved using machine learning techniques, allowing individuals to evaluate visual data and potentially react to our movements. This study investigates the development and implementation of machine learning methods for recognising hand movements in dynamic video data. This study builds on previous advances in machine learning and proposes a novel Sparse Gabor Descriptor (SGD)-based technique. Finally, these features can be classified using a random forest for gesture recognition. Experiments show that the suggested strategy delivers competitive results on gesture recognition datasets while requiring significantly less processing complexity. The comparison with roughly three state-of-the-art methodologies yields competitive results in terms of the system's many activities, such as accuracy and classification precision. The proposed technique achieves a recall rate of 99%, an accuracy rate of 94%, an F1-score of 97%, and a precision rate of 94%, which is higher than the KNN, NB, and LR methods.