15 lines
1.6 KiB
Plaintext
15 lines
1.6 KiB
Plaintext
- for participants P1 to P11, eyetracker had to be readjusted during the experiment so that pupil could still be tracked with high confidence. eyetracker was rarely readjusted between two recordings at the same depth, mostly it was in between two depths. as a consequence, usage of different depth calibrations for a test recording would have another source of error caused by slighly repositioning the cameras.
|
|
|
|
- p12 to p26 > no readjusting the tracker
|
|
- p23 > a minor adjustment for better pupil tracking confidence from 2nd depth
|
|
- p25 is myself. my head was slightly looking down or up in between recordings so even for a fixed depth the data could result in large errors! (not so useful). it's quite hard to do better all by myself.
|
|
|
|
- for p6, 001 is splited into 001 and 002 but the movement from test point 15 to 16 is lost, so from 001 only extact data for the first 15 points, and extract point 16 from the entire 002
|
|
- for p9, depth 0, calibration video, the screen is a bit outside of the screen from left, therefore at least two marker points are lost (cannot be tracked by ArUco). it might be better not to use the data from this depth or at least not use the left edge of the grids.
|
|
|
|
|
|
- p1 to p11 + p23 had minor adjustments to eye camera (and in terms scene camera) in between recordings
|
|
- p12 to p22 + p24 and p26 didn't have any camera adjustments in between recordings
|
|
- p25 is data recorded from myself
|
|
- p6 and p9 are special cases to be handled separately
|
|
- data for Hosna is not useful as the eye camera wasn't robustly tracking the eye (less than 20% frames with nonzero confidence) |