Calibrating gaze vector to Screen point ?

[ShreshthSaxena]

I'm trying to map the pred_gaze_np output to the 2D screen point. Is this something already implemented ? If not can you please help me with what approach should be followed.
I tried a simple polynomial regression from the vector to screen points and the results are decent but I'm wondering if there's a better approach.

[gv237-07]

Hi. Did you manage to figure it out please?

[ShreshthSaxena]

hey, yes I did... although this came out recently and it's really well documented here.

[gv237-07]

Thank you very much

[gkagkos]

@ShreshthSaxena @gv237-07 hey, I am also having troubles to do that and the link you have shared is down. Could you let me know what process you followed and how did you get the gaze vector from screen point? Thanks again!

[ShreshthSaxena]

hey @gkagkos You could map the vector to screen using geometric calibration for which you'll need the distance of user/eye from screen and the relative position of screen from camera. The paper from previous link explains it well : https://arxiv.org/abs/2104.12668
Another approach you can follow is to train a regression model using something like a 9-point calibration routine.