In response to the growing importance of AI-driven residential design and the lack of dedicated evaluation metrics, we propose the Residential Floor Plan Assessment (RFP-A), a comprehensive framework tailored to architectural evaluation. RFP-A consists of multiple metrics that assess key aspects of floor plans, including room count compliance, spatial connectivity, room locations, and geometric features. It incorporates both rule-based comparisons and graph-based analysis to ensure design requirements are met. A comparison of RFP-A and existing metrics was conducted both qualitatively and quantitatively, and it was revealed that RFP-A provides more robust, interpretable, and computationally efficient assessments of the accuracy and diversity of generated plans. We evaluated the performance of six existing floor plan generation models using RFP-A, showing that, surprisingly, only HouseDiffusion and FloorplanDiffusion achieved accuracies above 90%, while other models scored below or around 60%.
| Authors | Zeng, Pengyu and Yin, Jun and Gao, Yan and Li, Jizhizi and Jin, Zhanxiang and Lu, Shuai |
|---|---|
| Year | 2025 |
| Venue | Buildings (2075-5309) |
| DOI | 10.3390/buildings15101674 |
| Source Database | EBSCO (Applied Science & Technology) |
| Bridge-to-GNN | Category D |
| GNN Architecture | Graph GAN (heuristic — verify) |
| Graph Encoding | Room Adjacency Graph (inferred from title) |
| AEC Task | Floorplan / Layout Generation (inferred from title) |
| Cohort | Mature Applications (2024–2026) |
| Primary Metric | Accuracy (heuristic — verify) |
|---|---|
| Primary Value | 84.4% (heuristic — verify) |
| Key Finding | the Residential Floor Plan Assessment (RFP-A), a comprehensive framework tailored to architectural evaluation (heuristic — verify) |
| Quality Assessment | 2/6 · rigor: Low · Benchmark: ✓Baseline: ✓ (heuristic — verify) |
| GNN architectures detected | Graph GAN · Knowledge Graph Embedding (TransE/RotatE/DistMult) |
|---|---|
| Benchmark datasets referenced | PubMed · RPLAN |
| Dataset stats found | 25 edges · 70 edges |
| Reported metrics + values | Accuracy: 80% · Accuracy: 90% · Accuracy: 84.4% · Accuracy: 10% · Accuracy: 7% |
| Quality (heuristic, 6-flag) | 2/6 · rigor: Low Code/Data: —Benchmark: ✓Baseline: ✓CV/Split: —Ablation: —Reproducible: — |
Part of the GML/GNN in AEC Systematic Review (PRISMA 2020) — 112 papers, 2020–2026