These files contain the ground truth values of feature points for the sessions 08, 11, 17, and 21 for the Biwi kinect head pose dataset.

After converting the biwi recordings to videos every 30th frame of those recordings was hand labeled.
The frames might not exactly correspond to Biwi ones, due to indexing in Biwi dataset containing gaps (possibly due to frame drops).

As not all feature points are visible in every angle due to self occlusion we split the labels to 3 visibity cases, frontal, and both profiles.

The visibility indices are hence:
1 => [1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1];
2 => [0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 1 1 1 1 0 0 1 1 1 0 0 0 0 0 0 1 1 1 1 1 1 0 0 0 1 1 1 1 1 1 1 0 0 0 1 1 1 1 0];
3 => [1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 1 1 1 1 1 0 0 0 0 0 1 1 1 1 1 1 1 0 0 1 1 1 1 1 1 0 0 0 0 0 0 1 1 1 0 0 0 0 0 0 1 1 1 1 1 0 0 1 1];

The first line contains the visibility index (repeated twice for ease of reading into Matlab). 
Subsequent lines contain the feature point locations (x,y) in pixels.
The invisible pixels have junk values for the locations, that don't represent anything.