| Abstract [eng] |
This master’s thesis proposes an approach for Patients having neuro-motor disabilities who may occasionally lose their capacity to communicate with others as their condition deteriorates. This thesis presents a solution that makes it easier for these individuals to interact with others based on eye movements, supported by Google MediaPipe. Google MediaPipe has not been used so far by researchers for mouse interaction. Python was used for the coding, and Google Mediapipe, a framework for building machine learning pipelines for processing time series data. This cross-platform framework works on desktops, androids. This project tracked the landmarks of the eye. After getting the landmarks, it is used to move the cursor of the mouse using the Pyautogui module. Finally, the survey is conducted to determine the proposed system’s effectiveness. Furthermore, the test users tested the system in the absence of spectacles, and this system can detect the iris and pupil better than existing methods. The solution was compared with the existing solutions for mouse control. GPU/CPU speed (343.63 fps & 235.34 fps) and iris/pupil detection were both found to be superior to those of competing technologies. The average precision, however, was lower than that of the competing solutions. More research is needed to refine and enhance this system, and finally, some areas to consider in the future are presented. |
