Artificial Intelligence (AI) Empowered Sign Language Recognition Using Hybrid Neural Networks

Hand gestures serve as the primary means of communication in sign languages, which are composed of visual gestures made up of hands, faces, and other bodily motions. Although sign language has become more common in recent years, communicating with sign language speakers or signers remains difficult for non-sign language speakers. There has been promising progress in the disciplines of motion and gesture detection utilizing Artificial Intelligent techniques because of recent advances in deep learning and computer vision. The deep learning network makes full use of the advantages of time series classification provided by the recurrent neural network model as well as the feature extraction capabilities of convolutional neural networks to achieve more accurate recognition. High precision, scalability, and robustness, on the other hand, remain significant issues in sign language recognition research. The purpose of this research is to examine hybrid neural network to improve the accuracy and robustness of sign language recognition. This research proposes a sign language recognition system using an ensemble of convolutional neural network (CNN) models followed by a long short-term memory (LSTM) model. The proposed system is designed to recognize hand gestures and interpret sign language, with a focus on American Sign Language (ASL) and sign digit dataset. Research evaluated the performance of the proposed system using the same ASL dataset and achieved an accuracy of 99.3%. We compared the performance of the proposed model with standalone CNN models and found that the proposed hybrid model outperformed standalone models.