can count real valued loss values correctly.

to easily get a transformation matrix of a particular size.

these objects.

easy to get full_object_detections out of it.

here and there.

std::set is #included by the serialize header file.

both rectangle and full_object_detection objects.

legit feature vectors.

info.

they can support movable object part models.  Now all that needs to be
done is to implement the TODO inside the scan_image_pyramid object and
the movable part model support should be up and working.

cross_validate_graph_labeling_trainer() that can incorporate per node
loss values.