monocular depth estimation is often described as an ill-posed and inherently
ambiguous problem. Estimating depth from 2D images is a crucial step in scene
reconstruction, 3Dobject recognition, segmentation, and detection. The problem
can be framed as: given a single RGB image as input,