Abstract: In this paper, we propose a Transformer encoder-decoder architecture, called PoinTr, which reformulates point cloud completion as a set-to-set translation problem and employs a ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results