Speech Processing and Recognition of Sign Language using Tensor flow Classifier

Communication facilitates social interaction and perspective exchange. Therefore, the deaf and mute face significant obstacles when attempting to participate in society. People are only able to communicate through the use of sign language. It is possible to translate sign languages into a form that can be used for public communication. To this end, we will develop a real-time system that can interpret and translate Sign Language into written language. The majority of the operation relies on custom-made components such as gloves and robotic arms. To complete this task, we employ a convolutional neural network and a deep learning strategy. Beginning with the Python library Keras for training convolutional neural networks, a sign-based classifier model is developed. Then, we utilised the real-time skin segmentation system to locate the Region of Interest within the bounding box. The classified region is utilised to predict the symbol. The classifier was found to be more effective against complex backgrounds and camera angles, and to translate sign language into speech and text that the hearing-impaired can comprehend.