MetaSlot: Break Through the Fixed Number of Slots in Object-Centric Learning

2025-05-27

AI Summary

Existing object-centric learning (OCL) methods employ a fixed number of slots, leading to representation fragmentation—i.e., one object being assigned to multiple slots—in multi-object scenes. To address this, we propose MetaSlot, the first OCL framework to integrate vector quantization (VQ) codebooks into slot generation. MetaSlot dynamically adjusts the number of active slots via slot-level deduplication and progressive noise modulation, thereby overcoming the static capacity limitation inherent in Slot Attention. Crucially, it requires no prior knowledge of object count. Experiments demonstrate substantial improvements in object discovery and segmentation accuracy, achieving state-of-the-art performance on benchmarks including CLEVR and Multi-dSprites. Moreover, MetaSlot enhances object consistency and interpretability of slot representations, enabling more faithful and semantically coherent scene decomposition.

Technology Category

• Computer Vision: Representation Learning for Vision
• Search and Optimization: Learning to Search
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

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.

Problem

• Adapts Slot Attention to handle variable object counts
• Reduces duplicate slots via codebook quantization
• Improves aggregation stability with progressive noise injection

Innovation

• Adapts to variable object counts dynamically
• Uses codebook for object prototype quantization
• Injects noise to stabilize slot aggregation