1 paper
Caroline Chen, Sayna Ebrahimi, Fedor Kitashov +4
In this paper, we ask whether vision foundation models construct representations that reflect the intrinsic properties of 3D Euclidean space. Unlike previous works that probe 3D aw…