Object-Centric Slot Diffusion

TL;DR

This paper introduces Latent Slot Diffusion (LSD), a novel object-centric learning model that integrates diffusion models to improve scene understanding and generation without supervision, outperforming existing methods especially in complex scenes.

Contribution

LSD is the first object-centric model to replace slot decoders with a latent diffusion model conditioned on object slots, enabling unsupervised compositional generation without annotations.

Findings

LSD outperforms state-of-the-art transformer decoders in complex scenes.

LSD demonstrates superior unsupervised compositional generation quality.

Pre-trained diffusion models enhance LSD's real-world segmentation and generation capabilities.

Abstract

The recent success of transformer-based image generative models in object-centric learning highlights the importance of powerful image generators for handling complex scenes. However, despite the high expressiveness of diffusion models in image generation, their integration into object-centric learning remains largely unexplored in this domain. In this paper, we explore the feasibility and potential of integrating diffusion models into object-centric learning and investigate the pros and cons of this approach. We introduce Latent Slot Diffusion (LSD), a novel model that serves dual purposes: it is the first object-centric learning model to replace conventional slot decoders with a latent diffusion model conditioned on object slots, and it is also the first unsupervised compositional conditional diffusion model that operates without the need for supervised annotations like text. Through experiments on various object-centric tasks, including the first application of the FFHQ dataset in this field, we demonstrate that LSD significantly outperforms state-of-the-art transformer-based decoders, particularly in more complex scenes, and exhibits superior unsupervised compositional generation quality. In addition, we conduct a preliminary investigation into the integration of pre-trained diffusion models in LSD and demonstrate its effectiveness in real-world image segmentation and generation. Project page is available at https://latentslotdiffusion.github.io