Learning View Priors for Single-view 3D Reconstruction

Hiroharu Kato, Tatsuya Harada

There is some ambiguity in the 3D shape of an object when the number of observed views is small. Because of this ambiguity, although a 3D object reconstructor can be trained using…