Laptop Gaze Inference of dataset output about gazecapture HOT 6 CLOSED

Hi,

1. (3.3, 5.5) = user gazes 3.3 cm right and 5.5 cm up from the camera lens. We rotate the phones to get a larger screen estate. Just in your example we would never be able to show person a target above the camera lens if we did not rotate the device. It is only relevant for data collection, you do not need to worry about any rotations.
2. As long as your webcam is identical to the camera in our iPhone/iPad (some older versions of those) then it should work. You also need to either maintain the viewing distance or account for it. In your example you would only be able to track the gaze in the part of the screen that is covered by the ipad (15x20cm?). If you move away from the screen this scales up linearly (it is the viewing angle that matters at the end).

from gazecapture.

Hi, thanks for your answer. I still confused about the question 1.
1.1 so the camera point is the centre point(0,0) ? And does the camera is on the centre of the up bar of the phone/iPad?
1.2 let us image a common situation. we have just a single phone or laptop with a camera on its centre of up bar. as you means (3.3, 5.5) = user gazes 3.3 cm right and 5.5 cm up from the camera lens, is that means the gaze is out of the phone/laptop scale under this situation? And only when y < 0 the point is on the screen of the phone/laptop? Is it right?

from gazecapture.

1. Yes, camera is (0,0) regardless of device or its orientation.
2. Yes, in that case the user would be looking outside of the screen estate. And yes, for any screen point it would be y < 0.

from gazecapture.

Hi, so no matter where the camera is, the camera is (0,0) right?(because some laptop cameras are on the down bar of the screen(such as dell laptop)

from gazecapture.

Yes, (0,0) is always camera. There is no way to tell where the display is based on the camera image (= network input), so it cannot make any difference.

from gazecapture.

OK I get it. Thanks for your answer! really clear

from gazecapture.