Open-Universe Indoor Scene Generation using LLM Program Synthesis and Uncurated Object Databases

TL;DR

This paper introduces a novel indoor scene generation system that uses large language models to synthesize scene descriptions and retrieve unannotated 3D object meshes, enabling flexible, open-universe scene creation without extensive 3D training data.

Contribution

The system leverages LLMs for program synthesis and vision-language models for mesh retrieval, allowing open-universe indoor scene generation without large 3D datasets.

Findings

Outperforms traditional 3D generative models on scene quality.

Surpasses recent LLM-based layout methods in open-universe scenarios.

Effective retrieval of unannotated meshes from large databases.

Abstract

We present a system for generating indoor scenes in response to text prompts. The prompts are not limited to a fixed vocabulary of scene descriptions, and the objects in generated scenes are not restricted to a fixed set of object categories -- we call this setting indoor scene generation. Unlike most prior work on indoor scene generation, our system does not require a large training dataset of existing 3D scenes. Instead, it leverages the world knowledge encoded in pre-trained large language models (LLMs) to synthesize programs in a domain-specific layout language that describe objects and spatial relations between them. Executing such a program produces a specification of a constraint satisfaction problem, which the system solves using a gradient-based optimization scheme to produce object positions and orientations. To produce object geometry, the system retrieves 3D meshes from a database. Unlike prior work which uses databases of category-annotated, mutually-aligned meshes, we develop a pipeline using vision-language models (VLMs) to retrieve meshes from massive databases of un-annotated, inconsistently-aligned meshes. Experimental evaluations show that our system outperforms generative models trained on 3D data for traditional, closed-universe scene generation tasks; it also outperforms a recent LLM-based layout generation method on open-universe scene generation.