Empowering quadriplegic patients: A computer vision based approach for keyboard control using eye gaze and blink

Abstract

This study introduces a novel computer vision-based framework tailored to enhance text input efficiency for quadriplegia patients through their eye gaze and blink patterns. The proposed methodology capitalizes on the capabilities of OpenCV and the MediaPipe library within Python, facilitating accurate eye and iris detection. For streamlined communication, a personalized 2x4 grid keyboard layout is presented, enabling users to type by controlling an on-screen keyboard with their gaze and blinks. This arrangement optimizes typing speed while enabling grid navigation through gaze and selections by blinking. The system's effectiveness is gauged through a trial involving three volunteers, revealing significant typing speed enhancements as users familiarize themselves with the interface, culminating in an average Character Per Minute (CPM) of 14.53. Notably, the system stands out due to its cost-effective design, relying only on an economical webcam as the singular gaze detection method, thereby widening accessibility for a broader user spectrum. This inventive system holds the potential to revolutionize communication efficiency and intuitiveness for quadriplegia patients, furnishing them with a pragmatic avenue for text input and interaction.