Remote eye gaze tracker is a nonintrusive system which can give the coordinates of the position., where a person is looking on the screen. This paper gives an extensive analysis of a neural based eye gaze tracker. The eye gaze detection system based on neural network considers the variation of the system behavior with different feature extraction techniques adopted like eye template based features and features based on pupil detection. The performance comparison between these various models has been presented in this paper. The system has also been tested under different lighting conditions and distance from the webcam for different subjects. The performance of the eye gaze tracker based on features computed from the eyes were found to have better performance of 95.8% compared to the template based features.
H. Nandakumar and Amudha, J., “A comparative analysis of a neural-based remote eye gaze tracker”, in 2014 International Conference on Embedded Systems (ICES), 2014.