You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hello,
thank you for sharing the STG+ dataset. Currently I'm working on a similar gaze estimation task and therefore would like to use your dataset for training.
My question relates to the gaze_point entries in the (train|val)_meta.csv. They link to serialized numpy arrays and I'm able to load them as expected. But I have no clue what these values mean.
The values are two dimensional (xy or yaw,pitch or ...)
The values range from 3.11 to 56.66 on the first dimension and from 0.16 to 33.46 in the second dimension
Please help me clarify what these values are:
Are they coordinates on the screen in pixels (seems unlikely because of the value range)?
Are they coordinates on the screen in centimeters?
Are they angles in degrees (seems also unlikely because no negative values exist)?
In which coordinate system are they (camera, screen)?
Is there any possibility to calculate gaze angles from these values? (maybe using extrinsic camera parameters provided by this repository?)
Thanks in advance! :)
The text was updated successfully, but these errors were encountered:
Hello,
thank you for sharing the STG+ dataset. Currently I'm working on a similar gaze estimation task and therefore would like to use your dataset for training.
My question relates to the
gaze_point
entries in the(train|val)_meta.csv
. They link to serializednumpy
arrays and I'm able to load them as expected. But I have no clue what these values mean.3.11
to56.66
on the first dimension and from0.16
to33.46
in the second dimensionPlease help me clarify what these values are:
Thanks in advance! :)
The text was updated successfully, but these errors were encountered: