CAN: Communicator Assistance Network for Deaf-Dumb-Blind People

Communication is an integral part of our daily life for interacting with others. However, it is challenging for differently-abled people like deaf and/or dumb. Hence, they are bound to use sign language for essential communication. The sign languages are still not adequate for blind people; they use braille script to read and their voice for communication. This research paper aims to develop a neural network based Communicator Assistance Network (CAN) for the interaction among differently-abled people like blind and deaf and/or dumb. The proposed model detects the sign poses by a deaf and/or dumb person through a webcam and results the corresponding words, which have to be converted to speech for blind people; the whole process would be reversed for responding to the conversation. The proposed network has been trained for 15 epochs with batches of size 20 (taking 1459 steps to finish an epoch), and results in 98% accuracy. Furthermore, the well-trained model has been tested on real and unseen samples and it results in 100% accuracy. Therefore, the model is useful where the interaction with/between challenged people is more likely to happen, like in schools for specially challenged people or workplaces. This application is an asset to them in communicating with others and expressing their voice.