PolyLayout: Hierarchical VLM-Guided Layout Generation Beyond Rectangular Rooms

2026-08-11

AI Summary

This work addresses the limitations of existing indoor layout generation methods, which are predominantly confined to rectangular spaces and struggle to accommodate non-rectangular rooms and architectural constraints such as doors and windows commonly found in real-world residences, thereby hindering their practical deployment in retail applications. To overcome these challenges, the authors propose a hierarchical hybrid framework that, for the first time, integrates a vision-language model (VLM) into the non-rectangular layout generation pipeline. The approach operates through three synergistic stages: functional furniture clustering and intra-zone placement, VLM-guided macro-scale zoning of polygonal spaces, and rule-based collision-free micro-adjustments. Experiments on real irregular floor plans and large-scale product catalogs demonstrate that the proposed method significantly outperforms existing approaches in perceptual plausibility, geometric compliance, and inference efficiency.

Abstract

Generating physically plausible 3D room layouts is essential for home furnishing retail, enabling customers to visualize products in their own homes and confidently make purchasing decisions. However, a gap exists between academic research and real-world application: existing solutions primarily focus on algorithmic strategies for furniture placement, largely neglecting the non-rectangular geometries and strict door/window constraints prevalent in real homes. To bridge the gap, we introduce a hybrid, hierarchical framework tailored for retail, specifically designed to support scalable spatial planning applications. Our system decouples generation into three stages: (1) functional furniture clustering and fine-grained intra-zone placement; (2) macro-routing guided by a vision-language model (VLM) to anchor both these clustered zones and any remaining standalone furniture within diverse polygonal boundaries; and (3) rule-based optimization for collision-free micro-arrangements that respect architectural constraints. We evaluate our system on production-scale catalogs and a representative set of irregular real-world topologies. Our results show that our approach attains the highest perceptual plausibility while maintaining good geometric compliance at relatively low latency, and extends to irregular boundaries that existing methods do not natively support.

Problem

• 3D room layout generation
• non-rectangular rooms
• architectural constraints
• furniture placement
• real-world topologies

Innovation

• hierarchical layout generation
• vision-language model (VLM)
• non-rectangular room
• architectural constraints
• furniture clustering