we propose a general learning framework for joint 2D-3D scene understanding by selecting informative virtual 2D views of the underlying 3D scene. We feed both the 3D geometry and the generated virtual ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results