Soroor Aghapoor - Department of Electrical Engineering, Hamedan University of Technology, Hamedan, Iran
Mohammad Hossein Doost Mohammadi - Department of Electrical Engineering, Hamedan University of Technology, Hamedan, Iran

In this paper, a real-time gaze tracking algorithm with high accuracy is presented using the images obtained from an ordinary webcam. In this method, despite the methods that require special hardware setups, only low-resolution webcams are used, and the importance of this issue is that these webcams are available to the public. So, it can be used with no need to use special peripherals. As a result of simulation, a diagram has been obtained to determine the accuracy of the algorithm and has been investigated in different modes to achieve better accuracy. In this work, OpenCV and Dlib have been used to detect faces and eyes. Then, the pupil is detected by the processing of the eye structure, using some filters. So, the gaze is tracked in a real-time, high accurate scheme. Then, the gaze lines are drawn using the coordinates of tracking the gaze.

Gaze tracking, eye detection, Python