Interpolation based polynomial regression for eye gazing estimation: A comparative study

This paper presents a comparative study on the use of interpolation based gaze estimations with low-cost web camera with built in infrared LEDs. Six polynomial models were carefully compared based on at 9 points calibration; the best average accuracy from this study among all participants is found to be 1.10 degree from the center of gaze line. The best model provides better quality when compared to results from number of groups of research with web camera and standard open source gaze tracking software. In particular, the accuracy was acceptable under the reasonable price and very easy to establish. The results also emphasize the performance of the polynomial models that may vary with the environment and must be chosen carefully.