Gaze position detection by computing the three dimensional facial positions and motions

Gaze detection is to locate the position on a monitor screen where a user is looking. In our work, we implemented it by a computer vision system with a single camera above the monitor. We computed the initial three dimensional positions of facial features (both eyes, nostrils and lip comers) by a camera calibration and a parameter estimation algorithm, and also computed the moved 3D positions of those features by 3D rotation and translation estimations and affine transform. Finally, the gaze position on the monitor is computed from the normal vector of the plane determined by those moved 3D positions of features. Gaze detection error was 5.11 cm (RMS error) when the distance between the user and the monitor is 50-70 cm and a 19 in monitor is used.