WEBVTT

00:00:00.000 --> 00:00:02.910 align:middle line:84%
Why do we do calibration
of eye trackers?

00:00:02.910 --> 00:00:05.080 align:middle line:90%
Why it is so important?

00:00:05.080 --> 00:00:08.010 align:middle line:84%
Well, it's a necessary
step before any recording

00:00:08.010 --> 00:00:11.400 align:middle line:84%
that you do with eye
tracking for pupillometry,

00:00:11.400 --> 00:00:14.970 align:middle line:84%
because it establishes a
link between the position

00:00:14.970 --> 00:00:17.730 align:middle line:84%
of the pupil in the eye
and where in the world

00:00:17.730 --> 00:00:19.020 align:middle line:90%
the person is looking.

00:00:19.020 --> 00:00:23.910 align:middle line:84%
So what will happen if we
don't do the calibration?

00:00:23.910 --> 00:00:25.868 align:middle line:84%
Well, in many cases,
actually, the system

00:00:25.868 --> 00:00:27.660 align:middle line:84%
will not even allow
you to take a recording

00:00:27.660 --> 00:00:32.110 align:middle line:84%
because it's just not able to
record meaningful information.

00:00:32.110 --> 00:00:36.330 align:middle line:84%
But if the calibration
is not well done,

00:00:36.330 --> 00:00:40.657 align:middle line:84%
then you're just going to get
poor information about where

00:00:40.657 --> 00:00:41.490 align:middle line:90%
a person is looking.

00:00:41.490 --> 00:00:43.560 align:middle line:84%
So you think they're
looking in one spot,

00:00:43.560 --> 00:00:47.310 align:middle line:84%
and they're actually looking
in entirely different spots.

00:00:47.310 --> 00:00:49.980 align:middle line:84%
How do you calibrate a
stationary eye tracker?

00:00:49.980 --> 00:00:52.710 align:middle line:84%
Is there any specific
protocols we follow to do it?

00:00:52.710 --> 00:00:54.450 align:middle line:84%
Yeah, there's a pretty
standard protocol,

00:00:54.450 --> 00:00:56.485 align:middle line:90%
which we can demonstrate.

00:00:56.485 --> 00:00:57.735 align:middle line:90%
Would you like to have a seat?

00:00:57.735 --> 00:01:00.840 align:middle line:90%


00:01:00.840 --> 00:01:03.600 align:middle line:84%
So this is how we would
normally carry out

00:01:03.600 --> 00:01:06.780 align:middle line:84%
a calibration and recording
with the stationary eye tracker.

00:01:06.780 --> 00:01:10.440 align:middle line:84%
Here, we have, this is
the actual stationary eye

00:01:10.440 --> 00:01:12.870 align:middle line:84%
tracker, the camera
that's capturing his eyes,

00:01:12.870 --> 00:01:16.320 align:middle line:84%
and he's going to fix
his eyes on the screen.

00:01:16.320 --> 00:01:18.570 align:middle line:84%
And the software is
going to present him

00:01:18.570 --> 00:01:21.540 align:middle line:84%
with a series of visual
targets on the screen

00:01:21.540 --> 00:01:23.310 align:middle line:90%
in different locations.

00:01:23.310 --> 00:01:26.820 align:middle line:84%
And he's supposed to keep
his head still and follow

00:01:26.820 --> 00:01:29.350 align:middle line:90%
the targets with his eyes.

00:01:29.350 --> 00:01:30.555 align:middle line:90%
So we'll start this now.

00:01:30.555 --> 00:01:37.150 align:middle line:90%


00:01:37.150 --> 00:01:40.240 align:middle line:84%
And as this unfolds,
he should keep his eyes

00:01:40.240 --> 00:01:43.120 align:middle line:84%
as precisely as possible in
the middle of each target

00:01:43.120 --> 00:01:45.190 align:middle line:84%
and once we come
to the end of this,

00:01:45.190 --> 00:01:50.500 align:middle line:84%
now the system has a few samples
of positions of the pupil

00:01:50.500 --> 00:01:53.690 align:middle line:84%
and where that corresponded
to in the world.

00:01:53.690 --> 00:01:55.930 align:middle line:84%
So now it can generalise
from that information

00:01:55.930 --> 00:01:59.590 align:middle line:84%
and determine where
he's looking anywhere

00:01:59.590 --> 00:02:03.150 align:middle line:84%
on the screen on the
area that was calibrated.

00:02:03.150 --> 00:02:14.000 align:middle line:90%