Back to Wiki

[8] Graph2Plan: Learning Floorplan Generation from Layout Graphs

Core GNN/GML Structural Design & Layout Generation 2020

Paper Description extracted from PDF · intro excerpt · 146 words

HUANG∗, Shenzhen University Bedroom Bathroom Balcony 1+ 2 1 ... ... ... (a) Input building boundary. (b) Generated floorplans. (c) After adding room counts. (d) After adding room connectivity. (e) After layout graph editing. Kitchen Kitchen Living Room Living Room Living Room Living Room Living Room Living Room Living Room Bedroom BedroomBedroom Bedroom Bedroom Bedroom Br Br Br Br Bedroom Bedroom Bathroom Bathroom Bath Bath Bath Bath Balcony Balcony Bal Bal Kit Kit Balcony Balcony Bal. Bal. Kitchen Kitchen Kitchen Bedroom Bedroom Bedroom Bathroom Balcony Balcony Fig. 1. Our deep neural network Graph2Plan is a learning framework for automated floorplan generation from layout graphs. The trained network can generate floorplans based on an input building boundary only (a-b), like in previous works. In addition, we allow users to add a variety of constraints such as room counts (c), room connectivity (d), and other layout graph edits.

AuthorsHu, R.; Huang, Z.; Tang, Y.; Van Kaick, O.; Zhang, H.; Huang, H.
Year2020
VenueACM Transactions on Graphics 39(4), Art. 118
DOI10.1145/3386569.3392391
Source DatabaseBacktracking
Bridge-to-GNNCategory A
GNN ArchitectureGraph VAE (heuristic — verify)
Graph EncodingRoom Adjacency Graph (inferred from title)
AEC TaskFloorplan / Layout Generation (inferred from title)
CohortEarly Infrastructure (2020–2021)

Extracted Methodology heuristic — verify before citing

Primary MetricRecall (heuristic — verify)
Key Findingthe input boundary and aligned graph to the user in an interactive interface, where the user can edit the retrieved graph and adapt it as needed (heuristic — verify)
Quality Assessment3/6 · rigor: Medium ·
Benchmark: ✓Baseline: ✓Ablation: ✓ (heuristic — verify)

Detected Technical Stack auto-detected from PDF · regex catalogue match

GNN architectures detectedGraph VAE / VGAE · Knowledge Graph Embedding (TransE/RotatE/DistMult) · GCN
Benchmark datasets referencedRPLAN
Reported metrics + valuesIoU: 0.43 · IoU: 0.47 · IoU: 0.54 · IoU: 0.56 · IoU: 0.66
Quality (heuristic, 6-flag)3/6 · rigor: Medium
Code/Data: —Benchmark: ✓Baseline: ✓CV/Split: —Ablation: ✓Reproducible: —

Connected Papers Verification bibliometric (data/connected_papers_matches.json)

Seed of .bib export → verified neighboursThis paper was queried as a Connected Papers seed (41 entries scanned). In-corpus neighbours verified by the bibliometric ranking:

[37] 2020 · Huang · RECONCILING CITY MODELS WITH BIM IN KNOWLEDGE GRAPHS: A F…
[38] 2026 · Lu · Learning multi-dimensional sensor relationships for robus…
[65] 2021 · Quinn · A case study comparing the completeness and expressivenes…

Lineage Families

floorplan-generation

Enabled By / Precursors

[72] [73] [74] [75]

Extended By

[9] [10] [19] [88]

Infrastructure Dependencies

None

Research Frontier Tags

floorplan-generation

Part of the GML/GNN in AEC Systematic Review (PRISMA 2020) — 112 papers, 2020–2026