This method introduces an attention mechanism for object-centric learning of visual scenes, enabling models to decompose scenes into a set of distinct entities (or "slots"). 27.07.2023 17:54 aior