GSDiff: Synthesizing Vector Floorplans via Geometry-enhanced Structural Graph Generation

Sizhe Hu,Wenming Wu,Yuntao Wang,Benzhu Xu,Liping Zheng

2024-12-17

Abstract:Automating architectural floorplan design is vital for housing and interior design, offering a faster, cost-effective alternative to manual sketches by architects. However, existing methods, including rule-based and learning-based approaches, face challenges in design complexity and constrained generation with extensive post-processing, and tend to obvious geometric inconsistencies such as misalignment, overlap, and gaps. In this work, we propose a novel generative framework for vector floorplan design via structural graph generation, called GSDiff, focusing on wall junction generation and wall segment prediction to capture both geometric and semantic aspects of structural graphs. To improve the geometric rationality of generated structural graphs, we propose two innovative geometry enhancement methods. In wall junction generation, we propose a novel alignment loss function to improve geometric consistency. In wall segment prediction, we propose a random self-supervision method to enhance the model's perception of the overall geometric structure, thereby promoting the generation of reasonable geometric structures. Employing the diffusion model and the Transformer model, as well as the geometry enhancement strategies, our framework can generate wall junctions, wall segments and room polygons with structural and semantic information, resulting in structural graphs that accurately represent floorplans. Extensive experiments show that the proposed method surpasses existing techniques, enabling free generation and constrained generation, marking a shift towards structure generation in architectural design. Code and data are available at <a class="link-external link-https" href="https://github.com/SizheHu/GSDiff" rel="external noopener nofollow">this https URL</a>.

Graphics,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

This paper attempts to solve the problems of geometric inconsistency and generation complexity in the automatic design of building floor plans. Specifically, existing methods (including rule - based and learning - based methods) face challenges when dealing with design complexity and constraint generation, often requiring a large amount of post - processing and are prone to obvious geometric inconsistencies, such as misalignment, overlap and gaps.

To solve these problems, the paper proposes a new generation framework named GSDiff to synthesize vector floor plans through structure graph generation. GSDiff focuses on corner generation and wall segment prediction to capture the geometric and semantic aspects of the structure graph. To improve the geometric rationality of the generated structure graph, the authors propose two innovative geometric enhancement methods:

1. **Corner Generation**: A new alignment loss function is introduced to improve geometric consistency.
2. **Wall Segment Prediction**: A random self - supervised method is proposed to enhance the model's perception ability of the overall geometric structure, thereby promoting the generation of reasonable geometric structures.

By combining the diffusion model and the Transformer model as well as geometric enhancement strategies, GSDiff can generate corner, wall segment and room polygons containing structural and semantic information, thus accurately representing the floor plan. Extensive experiments show that this method outperforms the existing technology in all indicators, can achieve free generation and constraint generation, marking the transformation of building floor plan design towards structure generation.

### Key Contributions
1. **Novel Framework**: Transforms the vector floor plan generation into a structure graph generation process, thereby automatically generating diverse and high - quality vector floor plans.
2. **Alignment Error Optimization Strategy**: A new node alignment loss is introduced in the node generation stage, which improves the consistency of nodes.
3. **Edge - Aware Enhancement Strategy**: An innovative edge - aware enhancement strategy is introduced in the edge prediction stage, which improves the accuracy of edges.

These improvements make GSDiff more efficient, accurate in generating building floor plans and reduce the need for post - processing.