Watch Live Streaming on:
Beijing (GMT+8): 12:00 p.m., June 24
New York(EST): 12:00 a.m., June 24
Melbourne(GMT+10): 2:00 p.m., June 24
London(GMT): 05:00 a.m., June 24
A discussion on an urban layout workflow utilizing generative adversarial network (GAN) - With a focus on automatized labeling and dataset acquisition

Ximing Zhong
Ph.D. candidate at Aalto University | Architecture & Deep Learning | Human-machine collaboration for 3d space generation
Web: https://research.aalto.fi/en/persons/ximing-zhong
Ximing Zhong is a Ph.D. candidate at Aalto University, where he has been awarded a state public doctoral scholarship. His master school is Politecnico di Milano, Italy, and he is the recipient of a Global Gold Scholarship from Politecnico di Milano. His main research areas are deep learning, architectural design, graph neural networks GNN, GCN, Nerf-vae, 3D reconstruction, human-machine fusion architectural decision making, reinforcement learning to meet architect preferences, environmental perception of complex spaces, and design intuition machine learning.
In addition, he is a lecturer in the Introduction to Artificial Intelligence Assisted Urban Design Methods course (course code A010125) at Xi'an University of Architecture and Technology, a registered architect in Finland (SAFA), and one of the principals of Archiford Associates, whose official website is http://www.archiford.com, and who leads most of the projects above.
Abstract
Deep Learning (DL) has recently gained widespread attention in the automation of urban layout processes. This study proposes a rule-based and Generative Adversarial Network (GAN) workflow to automatically select and label urban datasets to train customized GAN models for the generation of urban layout proposals. The developed workflow automatically collects and labels urban typology samples from open-source maps. Furthermore, it controls the results of the GAN process with labels and provides real-time urban layout suggestions based on a co-design process. The conducted case study shows that the average value of the GAN results, trained from an automatically generated dataset, meets the site's requirements. The developed co-design strategy allows the architect to control the GAN process and perform iterations on urban layouts. The research addresses the research gap in GAN applications in the field of urban design and planning. Many studies have demonstrated that training the (GAN) model by labeling enables machines to learn urban morphological features and urban layout logic. However, two research gaps remain: (1) The manual filtering of GAN urban sample datasets to fit site-specific design requirements is very time-consuming. (2) Without a suitable data labeling method, it is difficult to manage the GAN process in such a manner to facilitate the meeting of overriding design requirements.

Host
Wei Wu
Wei Wu is a designer and computational artist with a Master's degree in Design Studies from Harvard University Graduate School of Design. She operates at the intersection of design and emerging technologies, producing work that encompasses robotic installations, interactive media art, and extended reality design.
Commentaires