Using Ignorance in 3D Scene Understanding

Figure 1

Assumed visible (green) and occluded (red) regions of a hypothetical object (a chair) obtained by spatial reasoning (output on the depth map).