Is there anybody knowing opencv well who would be interested
implementing eye tracking to find out how accurately a mouse could be
driven by eyes this way?
I imagine the following features:
1) assume head position is fixed (for simplicity)
2) show 16 points on screen and measure position of eyes
3) interpolate mouse position to position on screens setting the mouse
pointer and or show target points of both eyes separately in a way.
Is there anybody who can tell me whether opencv is a good tool to
implement such prototype and how much work it would be? If somebody is
interested in giving it a try I could also pay for it rather than
learning about all details - drop me a private mail then.
I'm fine with publishing the results - code as well - no problem.
If anybody of you have more experience with eye tracking being able to
tell me about its limits I'd be very interested to hear about your
experiences as well. I know about tobii.com and similar - but because
I'm wearing glasses I expect the results to be accurate enough to hit a
window, maybe a 200x200 px area, but maybe not accurate enough to place
a cursor at the beginning of a word (?). If a glasses are a problem VR
could be used .. there are many ways.
I imagine that you can move eyes very fast and that it would be more
confident than the 'where is the mouse, where do I want it to be, how to
move it there' feedback loop in the head when clicking so I want to