Depth Field Networks for Generalizable Multi-view Scene Representation

Abstract

Fig. 1: Our Depth Field Networks (DeFiNe) achieve state of the art in multiview depth estimation, while also enabling predictions from arbitrary viewpoints.