Development of a Sign Language for Total Paralysis and Interpretation using Deep Learning

People with total paralysis suffer not just from physical disability but also from the agony of the inability to communicate and express their feelings. Researches have long tried to solve this problem with various techniques such as collecting the patient's message directly from the brain using Brain-Computer interface and determining the gaze of the patient on a screen that contains letters and symbols. The first approach proved to be expensive and less accurate. The second one needs very fine calculation of the pupil center and even if done correctly the drifting of the human eye makes it hard to get a better accuracy. Again, all of these methods are dependent on electronic devices. In this paper, we proposed a solution that does not require the help of any electronic devices by developing a sign language for eye movements. We also proposed an approach to automatically interpret the patient's sign using Convolutional Neural Network and achieved great accuracy on the training data. This suggests that our network can be used to provide a better communication method for paralyzed patients.