LoCUS: Learning Multiscale 3D-consistent Features from Posed Images

Open in new window