Introducing DOCTR, a novel object-centric Transformer for unified learning with multiple objects in point scene understanding.
