Advanced Machine Learning and Deep Learning Techniques for Gujarati Sign Language Recognition

Recognizing sign language is essential for helping people who are deaf or hard of hearing communicate, allowing for smooth engagement in a variety of Social, Educational and Professional settings. Despite its significance, the limited understanding of sign language by the broader population presents challenges in accessibility and inclusivity. Recent developments in AI and ML have cleared the path for fully automated systems that can translate gestures into text or speech with high accuracy. This research focuses on Gujarati sign language, specifically recognizing 34 consonant letters. Using a dataset of 3,798 training images and 749 testing images, several machine learning models were evaluated, including Support Vector Machines (SVM), Logistic Regression (LR), Random Forest, Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), ResNet50, and VGG19. Among these, VGG19 achieved the highest accuracy of 94%. Building on this success, the system will be extended to include commonly used words, ensuring practical application for real-world communication. This work highlights the transformative potential of AI-driven solutions in enhancing the ability to communicate for those who are hard of hearing, fostering inclusivity, and addressing societal communication challenges.