Assistive Technology to communicate through eye blinks - A Deep learning approach

Abstract

Paralysis is the loss of muscle control in parts of or the whole body. Only 15% of people affected by paralysis are employed, due to concerns over their ability to function properly. These patients are able to think and have ideas just as well as other people but are unable to communicate these thoughts to their full extent because of a lack of an affordable medium of communication. In this paper, we propose a system for paralysis patients to communicate through Morse encoded eye blinks which can be decoded through any device equipped with a webcam. It provides a novel, affordable and practical way of communicating the entire English vocabulary along with a smaller learning curve, which is not provided in other methods. The detected pattern of blinks is converted into text for human understanding. It is compared with current AAC (Augmentative and Alternative Communication) devices and standard blink detection techniques and the results are provided for the deep learning solution on a standard dataset.