Published Paper Details:

Through communication, people can engage and share thoughts and feelings. There are several obstacles in the way of the deaf community's social interactions. The individuals use sign language to communicate with one other. In order to communicate with regular people, a technology can convert sign languages into a form that is understandable. The goal of this project is to create a real-time text-to-Indian Sign Language (ISL) translation system. Most of the work is done by hand. In this paper, we present a deep learning technique for classifying signs using a convolutional neural network. Using the numerical signs and the Python-based Keras convolutional neural network implementation, we first build a classifier model. Phase two involved using a second real-time system that located the Region of Interest in the frame that displays the bounding box using skin segmentation. The segmented region is fed into the classifier model in order to forecast the sign. For the identical subject, the system's accuracy rate is 99.56%; in low light, it is 97.26%. The classifier was seen to be becoming better with different background and angle of image capture. Our approach focuses on the RGB camera system.