Scene Representation Networks: Continuous 3D-Structure-Aware Neural Scene Representations

Vincent Sitzmann, Michael Zollhofer, Gordon Wetzstein, NIPS-2019


The paper proposes a novel technique for 3D scene representations trained end-to-end from only 2D images and their camera poses, without access to depth or shape. For better results than the previous contributions in the field the authors also propose a novel ray-marching algorithm for rendering. This paper won Honorable Mention Outstanding New Directions Paper Award at NeurIPS 2019.

Main Contributions

Some other details

Our two cents