3D Reconstruction of non-visible surfaces of objects from a Single Depth View -- Comparative Study
This addresses the problem of object reconstruction for robotics applications like collision-free planning, but it is incremental as it compares existing methods.
The paper compared two methods, DeepSDF and MirrorNet, for reconstructing non-visible surfaces of objects from a single depth view, finding that MirrorNet is faster and has smaller reconstruction errors in most categories.
Scene and object reconstruction is an important problem in robotics, in particular in planning collision-free trajectories or in object manipulation. This paper compares two strategies for the reconstruction of nonvisible parts of the object surface from a single RGB-D camera view. The first method, named DeepSDF predicts the Signed Distance Transform to the object surface for a given point in 3D space. The second method, named MirrorNet reconstructs the occluded objects' parts by generating images from the other side of the observed object. Experiments performed with objects from the ShapeNet dataset, show that the view-dependent MirrorNet is faster and has smaller reconstruction errors in most categories.