SPACE: Unsupervised Object-Oriented Scene Representation via Spatial Attention and Decomposition

TL;DR

SPACE is a generative model that unifies spatial attention and scene mixture techniques to efficiently decompose complex scenes into object and background representations, scalable to many objects.

Contribution

It introduces a probabilistic framework that combines the strengths of previous methods, enabling scalable, detailed scene decomposition with explicit object and background representations.

Findings

Outperforms previous models like SPAIR, IODINE, and GENESIS.

Effectively decomposes scenes with many objects without performance loss.

Demonstrates applicability on Atari and 3D-Rooms datasets.

Abstract

The ability to decompose complex multi-object scenes into meaningful abstractions like objects is fundamental to achieve higher-level cognition. Previous approaches for unsupervised object-oriented scene representation learning are either based on spatial-attention or scene-mixture approaches and limited in scalability which is a main obstacle towards modeling real-world scenes. In this paper, we propose a generative latent variable model, called SPACE, that provides a unified probabilistic modeling framework that combines the best of spatial-attention and scene-mixture approaches. SPACE can explicitly provide factorized object representations for foreground objects while also decomposing background segments of complex morphology. Previous models are good at either of these, but not both. SPACE also resolves the scalability problems of previous methods by incorporating parallel spatial-attention and thus is applicable to scenes with a large number of objects without performance degradations. We show through experiments on Atari and 3D-Rooms that SPACE achieves the above properties consistently in comparison to SPAIR, IODINE, and GENESIS. Results of our experiments can be found on our project website: https://sites.google.com/view/space-project-page