Low-cost Remote Eyetracking Accuracy Test

[oVISIONo]

Hi Walter, one more question...

Do you think I could get better results tracking both eyes than just one?.

Thanks.

[piechulla1966]

Maybe. I'm sure I don't know.

[oVISIONo]

no problem...

Another question. Do you use the homography matrix to calculate the gaze point?

thanks.

[piechulla1966]

No, I do not. BUT: "All mappings provide reasonable accuracy. However, the second-order mapping and homographic mappings result in the best performance." (Denghong Li, 2006,p. 43).

Li, D. (2006). Low-cost eye-tracking for human computer interaction. Masters Thesis, Iowa State University Human Computer Interaction Program.

Please find this valuable document on the web if you havn't already got it.

[oVISIONo]

Thanks for your answer Walter.

I will check it out.

[ak2agent26]

@oVISIONo yeah i read in some paper that it reduces the inaccuracy of by averages

[oVISIONo]

Hi Walter,

Could I ask you what software do you use to capture all that screens in one? thanks.

[piechulla1966]

Only the left-top quadrant contains what is captured. The rest just helps me to see what's going on in different levels of processing. See: OpenCV library wiki --> Frequently Asked Questions --> How to display more than one image in a single window?

Cheers, have fun!

[oVISIONo]

Thanks a lot for your answer! That's what I was looking for...

[oVISIONo]

Hi Walter,

I'd like to know how do you keep track of the gaze direction besides of the head movements...

[piechulla1966]

See Helmholtz's (1925) treatise on physiological optics (Vol. II). Or watch the video carefully and think it over. You may also google for "corneal reflex". Hope this helps getting you started. Cheers, have fun!

[oVISIONo]

Thanks Walter. I will check it out.