Binocular Gaze Estimation Method

[yang-ruixin]

Dear authors,

Thanks for your great work.
In terms of binocular gaze estimation, you cite Lech Swirski: "Gaze estimation on glasses-based stereoscopic displays", Chapter: 7.4.2 Cyclopean gaze estimate.
But in Lech's thesis, it seems there are markers to the back of the eye cameras and there is another camera towards the markers to record the position of the eye camera.
But for Pupil hardware, we don't have the markers and this additional camera.
Where did I miss?

Thanks in advance.

[Pablo Prietz]

Hi @yang-ruixin

We cite Swirski here because we apply their approach of intersecting two monocular gaze rays to determine a cyclopean gaze ray.

We estimate the eye-to-scene-camera poses based on bundle adjustment, using the collected data during calibration. Therefore, there is no need for markers or external cameras.