Eye-tracking on Smartphones using Regression-based Prediction

In this paper, we propose a regression-based prediction algorithm that can be used in eye-tracking and/or blinking detection systems on smartphones with accelerometers. We use linear regression as an instrument to calibrate accelerometers in our eye-location prediction algorithm. Eye tracking and/or blinking detection algorithms have various applications in mobile environments. For example, to mitigate Computer Vision Syndrome (CVS) from which many smartphone users suffer, eye blinking detection systems running on smartphones can give advices to users regarding their eye blinking habits. Experimental results on real smartphones show that our regression based prediction shows around 30 percent points improvement over a nave implementation of an existing solution.