TY - GEN
T1 - Comparison of GAN-based Spatial Layout Generation
T2 - 42nd Conference on Education and Research in Computer Aided Architectural Design in Europe, eCAADe 2024
AU - Park, Hyejin
AU - Gu, Hyeongrno
AU - Hong, Soonmin
AU - Choo, Seungyeon
N1 - Publisher Copyright:
© 2024, Education and research in Computer Aided Architectural Design in Europe. All rights reserved.
PY - 2024
Y1 - 2024
N2 - Recent advancements in Large Language Models (LLM) andthe emergence of ChatGPT are rapidly progressing Generative AI models, suggesting the possibility of AI replacing human creative activities. In architecture, -wheire outcomes depend on human creative thinking, the pre-planning stage is crucial.Architecturalplanning 5 decisions on mass, space layout, and space program, aimingfor optimal design with a significant impact on subsequent stages. Creating a client-centric design within a given time prompts architects to searchfor diverse reference materials. However’finding comparable spatial layouts is challenging due to the predominant focus on materials, construction methods, and details. This study introduces AIBIM-Spacemaker, a GenerativeAdversaria! Network (GAN)-based program -we developed for generating spatial layouts through graphical composition of space programs· Focusing on a house with limited space usage but versatile layouts, the study collected 10,000 raster-basedfloor plan images, creating a training dataset annotatedfor spatial elements. Training this dataset using the YOLO model enabled automatic extraction of vector-based data representing spatial relationships from raster-based images. .A GAN trained on this data resulted in ÄIBIM- Spacemaker, allowing users to create diverse spatial layouts. Executing a graph with nodes representing spaces and edges denoting relationships between doors and windows using the trained GAN produced varied spatial layouts. Verification, comparing actual ground truth values, GAN-generated outcomes, and architect-provided values confirmed the program's effectiveness in the planning stage. Performance ^as verified by comparing the program, learning method, dataset, and results developed in this study with previous studies on GAN-based spatial layout generation. This study identifies thepotential for AI- based spatial layout generation, enhancing planning efficiency and contributing to intelligent design automation, with anticipated positive impacts on planning task efficiency.
AB - Recent advancements in Large Language Models (LLM) andthe emergence of ChatGPT are rapidly progressing Generative AI models, suggesting the possibility of AI replacing human creative activities. In architecture, -wheire outcomes depend on human creative thinking, the pre-planning stage is crucial.Architecturalplanning 5 decisions on mass, space layout, and space program, aimingfor optimal design with a significant impact on subsequent stages. Creating a client-centric design within a given time prompts architects to searchfor diverse reference materials. However’finding comparable spatial layouts is challenging due to the predominant focus on materials, construction methods, and details. This study introduces AIBIM-Spacemaker, a GenerativeAdversaria! Network (GAN)-based program -we developed for generating spatial layouts through graphical composition of space programs· Focusing on a house with limited space usage but versatile layouts, the study collected 10,000 raster-basedfloor plan images, creating a training dataset annotatedfor spatial elements. Training this dataset using the YOLO model enabled automatic extraction of vector-based data representing spatial relationships from raster-based images. .A GAN trained on this data resulted in ÄIBIM- Spacemaker, allowing users to create diverse spatial layouts. Executing a graph with nodes representing spaces and edges denoting relationships between doors and windows using the trained GAN produced varied spatial layouts. Verification, comparing actual ground truth values, GAN-generated outcomes, and architect-provided values confirmed the program's effectiveness in the planning stage. Performance ^as verified by comparing the program, learning method, dataset, and results developed in this study with previous studies on GAN-based spatial layout generation. This study identifies thepotential for AI- based spatial layout generation, enhancing planning efficiency and contributing to intelligent design automation, with anticipated positive impacts on planning task efficiency.
KW - Generative Adversaria! Networks(GNN)
KW - Pre-design stage
KW - Space Layout Generation
KW - Space Program
KW - You OnfyLook OncefFOLO)
UR - http://www.scopus.com/inward/record.url?scp=85209801514&partnerID=8YFLogxK
U2 - 10.52842/conf.ecaade.2024.1.539
DO - 10.52842/conf.ecaade.2024.1.539
M3 - Conference contribution
AN - SCOPUS:85209801514
SN - 9789491207372
T3 - Proceedings of the International Conference on Education and Research in Computer Aided Architectural Design in Europe
SP - 539
EP - 548
BT - Proceedings of the 42nd Conference on Education and Research in Computer Aided Architectural Design in Europe, eCAADe 2024
A2 - Kontovourkis, Odysseas
A2 - Phocas, Marios C.
A2 - Wurzer, Gabriel
PB - Education and research in Computer Aided Architectural Design in Europe
Y2 - 9 September 2024 through 13 September 2024
ER -