MetaSlot: Break Through the Fixed Number of Slots in Object-Centric Learning
This addresses a bottleneck in object-centric representation learning for computer vision, enabling better generalization in next-generation Pre-trained Vision Models, though it is an incremental improvement over existing Slot Attention variants.
The paper tackles the problem of fixed slot counts in Object-Centric Learning, which causes objects to be split into parts when object numbers vary, by introducing MetaSlot, a plug-and-play variant that adapts to variable object counts, achieving significant performance gains and more interpretable representations across multiple datasets and tasks.
Learning object-level, structured representations is widely regarded as a key to better generalization in vision and underpins the design of next-generation Pre-trained Vision Models (PVMs). Mainstream Object-Centric Learning (OCL) methods adopt Slot Attention or its variants to iteratively aggregate objects' super-pixels into a fixed set of query feature vectors, termed slots. However, their reliance on a static slot count leads to an object being represented as multiple parts when the number of objects varies. We introduce MetaSlot, a plug-and-play Slot Attention variant that adapts to variable object counts. MetaSlot (i) maintains a codebook that holds prototypes of objects in a dataset by vector-quantizing the resulting slot representations; (ii) removes duplicate slots from the traditionally aggregated slots by quantizing them with the codebook; and (iii) injects progressively weaker noise into the Slot Attention iterations to accelerate and stabilize the aggregation. MetaSlot is a general Slot Attention variant that can be seamlessly integrated into existing OCL architectures. Across multiple public datasets and tasks--including object discovery and recognition--models equipped with MetaSlot achieve significant performance gains and markedly interpretable slot representations, compared with existing Slot Attention variants.