I used the Stream Recorder app (in HoloLens2ForCV) to capture hand tracking and eye tracking data, and exported data like "XXX_head_hand_eye.csv"
But the problem I'm having is that the file I'm exporting contains hundreds of columns of data (data from ColumnA to ColumnAGC) and I don't know how to match what you've listed.
Let me describe roughly what I am exporting:
Here are the contents of each column
A:Timestamp
B~M: some floats that might be coordinate axes
N~PS: a lot of boolean values, containing a lot of columns of zeros.
PT~QE: some floats that may be coordinates of axes
QF~QI: boolean values, each row is 0, 0, 0, 1
QJ~QU: some floats that may be coordinates. (the first 7 columns are the same as the first 7 columns of PT~QE)
QV~QY: boolean values, each row is 0, 0, 0, 1
QZ~RK: some floats that may be coordinates of the axis, different from the numbers in the previous columns.
RL~RO: boolean values, each row is 0, 0, 0, 1
Next, as in QZ~RO, all the way to AFS is a combination of 12 columns of float + 4 columns of boolean (0, 0, 0, 1).
Then the contents of AFT~AGB are all zeros. Finally, there is a column of float, and every number is 0.001709.
In the absence of column headings, other than timestamps, I'm having trouble pairing these columns well with the head, hand joint, and eye tracking data you mentioned.