Inside Google Street View, moving the mouse around over different buildings and stuff, it highlights the 3D plane in which that surface of the building is located.
How does it recognize this thing? Is it done automatically by machine algorithms, or manually in the preprocessing?