Abstract
Appearance-based gaze estimation aims to directly learn a mapping from face images to gaze directions. Via the front camera of mobile phones or tablets, face and eye images are obtained and used to extract Convolutional Neural Net-work(CNN) features. As a natural human computer interaction method, gaze is very efficient and has potential to be widely used on mobile phones and tablets. Grid layout is popular and commonly used in mobile applications user interface(UI) design and interaction. In this study, we propose a novel interaction method based on gaze estimation-GazeGrid. We conduct experiments on a large-scale dataset. The proposed method achieves more than 86% and 93% accuracy rates on mobile phones and tablets. Further more, we provide optimal GazeGrid sizes for 11 types of mobile devices based on experimental results. We believe that GazeGrid can help to accelerate gaze driven applications and interactions.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
