DStruct2Design: Data and Benchmarks for Data Structure Driven Generative Floor Plan Design

TL;DR

DS2D tackles data-structure driven floorplan design by unifying existing datasets into a JSON-based representation that encodes room polygons and numerical constraints. It introduces a dataset and benchmarks, including self-consistency, prompt-consistency, and GED-based compatibility, and demonstrates feasibility by fine-tuning LLaMA3-8B-Instruct-based models on the task. The results show the approach can achieve strong numerical adherence and reasonable compatibility with input constraints, while revealing limitations such as room overlaps that warrant further methodological improvements. Overall, this work provides a practical path to constraint-aware, geometry-preserving floorplan generation and fosters future research on richer data representations and evaluation metrics.

Abstract

Text conditioned generative models for images have yielded impressive results. Text conditioned floorplan generation as a special type of raster image generation task also received particular attention. However there are many use cases in floorpla generation where numerical properties of the generated result are more important than the aesthetics. For instance, one might want to specify sizes for certain rooms in a floorplan and compare the generated floorplan with given specifications Current approaches, datasets and commonly used evaluations do not support these kinds of constraints. As such, an attractive strategy is to generate an intermediate data structure that contains numerical properties of a floorplan which can be used to generate the final floorplan image. To explore this setting we (1) construct a new dataset for this data-structure to data-structure formulation of floorplan generation using two popular image based floorplan datasets RPLAN and ProcTHOR-10k, and provide the tools to convert further procedurally generated ProcTHOR floorplan data into our format. (2) We explore the task of floorplan generation given a partial or complete set of constraints and we design a series of metrics and benchmarks to enable evaluating how well samples generated from models respect the constraints. (3) We create multiple baselines by finetuning a large language model (LLM), Llama3, and demonstrate the feasibility of using floorplan data structure conditioned LLMs for the problem of floorplan generation respecting numerical constraints. We hope that our new datasets and benchmarks will encourage further research on different ways to improve the performance of LLMs and other generative modelling techniques for generating designs where quantitative constraints are only partially specified, but must be respected.

Figures (3)

• Figure 1: Model Pipeline Overview
• Figure 2: Example generations and extracted bubble diagrams. Pink background: ground truth floor plan and its associated bubble diagram. Green background: from F+BD model with Specific prompts. Blue background: from F+BD model with Total Area prompts.
• Figure 3: This Floor plan is generated by the string output below