InHabit: Leveraging Image Foundation Models for Scalable 3D Human Placement

2026-04-21

AI Summary

The scarcity of large-scale, semantically plausible, and contextually consistent 3D human-scene interaction data hinders embodied agents’ understanding of 3D environments. This work proposes InHabit, the first framework to transfer knowledge from internet-scale 2D vision foundation models to the task of 3D human placement, establishing an end-to-end automated generation pipeline. It leverages a vision-language model to recommend contextually appropriate actions, employs an image editing model to synthesize humans, and refines their poses into SMPL-X representations aligned with scene geometry. Built upon Habitat-Matterport3D, the method produces a large-scale dataset comprising 78K samples across 800 building-scale scenes, significantly improving performance in RGB-based human reconstruction and contact estimation. In user studies, 78% of participants preferred its outputs over those of existing methods.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Computer Vision: Large Vision Models
• Intelligent Robots: Embodied AI

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience

Abstract

Training embodied agents to understand 3D scenes as humans do requires large-scale data of people meaningfully interacting with diverse environments, yet such data is scarce. Real-world motion capture is costly and limited to controlled settings, while existing synthetic datasets rely on simple geometric heuristics that ignore rich scene context. In contrast, 2D foundation models trained on internet-scale data have implicitly acquired commonsense knowledge of human-environment interactions. To transfer this knowledge into 3D, we introduce InHabit, a fully automatic and scalable data generator for populating 3D scenes with interacting humans. InHabit follows a render-generate-lift principle: given a rendered 3D scene, a vision-language model proposes contextually meaningful actions, an image-editing model inserts a human, and an optimization procedure lifts the edited result into physically plausible SMPL-X bodies aligned with the scene geometry. Applied to Habitat-Matterport3D, InHabit produces the first large-scale photorealistic 3D human-scene interaction dataset, containing 78K samples across 800 building-scale scenes with complete 3D geometry, SMPL-X bodies, and RGB images. Augmenting standard training data with our samples improves RGB-based 3D human-scene reconstruction and contact estimation, and in a perceptual user study our data is preferred in 78% of cases over the state of the art.

Problem

• 3D human-scene interaction
• embodied agents
• synthetic data generation
• scene context
• human placement

Innovation

• foundation models
• 3D human-scene interaction
• SMPL-X
• image-to-3D lifting
• scalable data generation