The text format of this data may be easier to load, but will probably be much slower.  If this is annoying, use the binary format.


Each sequence dir contains images, raw depth data, and initial segmentations of objects to track and segment.

For example:

sequence06/1288573043.376525.jpg - image
sequence06/1288573043.376525.txt - depth data
sequence06/1288573043.376525_segmentation.txt - segmentation

Depth data contains one point per row.  Each row is:

x y z x' y' z' intensity u v

where x, y, and z are the points in a global coordinate system, x', y', and z' are the points in the depth sensor coordinate system, intensity is the laser return intensity, and u v are the image coordinates of the point.


Segmentations are formatted as follows:

number of segmented objects
object id
number of points
point
point
...
object id
number of points
point
point
...
...

and so on.  Each "point" above is an index into a row of the depth data for the scene.