GAN as a generative architectural plan layout tool: A case study for training DCGAN with Palladian Plans and evaluation of DCGAN outputs

Uzun, Can; Çolakoğlu, Meryem Birgül; İnceoğlu, Arda

doi:10.5505/itujfa.2020.54037

Cited by 11 publications

(6 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Various housing layout design assistant approaches and GAN-based applications, such as DCGAN (Uzun et al, 2020), ActFloor-GAN (Wang et al, 2023), ArchiGAN (Chaillou, 2020), and HouseGAN (Nauata et al, 2020), exist in the field. However, a notable limitation of these tools is the absence of user interfaces to facilitate experimental application.…”

Section: Selected Design Assistant Toolsmentioning

confidence: 99%

The Role of AI Design Assistance on the Architectural Design Process: An Empirical Research with Novice Designers

Zeytin,

Öztürk Kösenciğ,

Öner

2024

Journal of Computational Design

View full text Add to dashboard Cite

This study explores the integration of Generative Design Assistants (GDAs), specifically machine learning based tools, in the architectural design process. It investigates how these tools, once confined to experimental realms, are now influencing mainstream architectural practice, particularly among novice architects. The research focuses on third and fourth-year architecture students, examining how they adapt to and integrate these advanced AI tools into their design workflows. Through an empirical online workshop, the study collected data of design process recordings, design output success scores of students by an independent jury, and post-experiment surveys. This approach provided insights into the timing, frequency, and sequence of GDA usage, as well as the influence of specific GDA features on design success. The research reveals that three primary strategies emerged in students' GDA usage: continuous use throughout the design process, selective problem-solving use, and initial ideation use followed by traditional methods. However, an over-reliance on GDAs was noted to potentially limit the designer’s interpretive and developmental input. The survey shows that different GDAs have distinct strengths and impacts on the design process. In terms of selected GDAs for the experiment, ArchiGAN aids in discovery and ideation, while HouseGAN excels in reframing design problems. In conclusion, the study underscores the transformative potential and challenges of GDAs in architectural design and highlights the need for balanced GDA integration. The research outputs show that future research should focus on the long-term implications of GDAs in architectural education. This research aims to guide the effective integration of AI in architecture, enhancing the human designer's role rather than overshadowing it.

show abstract

Section: Selected Design Assistant Toolsmentioning

confidence: 99%

The Role of AI Design Assistance on the Architectural Design Process: An Empirical Research with Novice Designers

Zeytin,

Öztürk Kösenciğ,

Öner

2024

Journal of Computational Design

View full text Add to dashboard Cite

show abstract

“…Dengan cara yang sama, dia membawa Artificiaal Intelligence selangkah lebih dekat ke dalam dunia arsitektur: menggambar dan produksi gambar. Secara keseluruhan, dari jaringan sederhana hingga GAN, alat generasi baru yang digabungkan dengan daya komputasi yang semakin murah dan dapat diakses kini memposisikan Artificiaal Intelligence sebagai media yang terjangkau dan kuat [21]. Gambar 6.…”

Section: Artificial Intelligence Sebagai Pendekatan Baruunclassified

Peran Artifical Intelligence dalam Tahap Perencanaan dan Perancangan Desain Arsitektur

Effendi

Satwiko

2021

Journal of Digital Architecture

View full text Add to dashboard Cite

Perkembangan teknologi informasi di masa modern ini menyetuh setiap sendi-sendi kehidupan yang membuat segalanya menjadi lebih mudah dan praktis. Situasi ini juga menggambarkan apa yang terjadi dalam bidang arsitektur yang menuntut peningkatan kompleksitas desain yang lenih tinggi dengan waktu yang lebih cepat dari masa-masa sebelumnya. Penggunaan teknologi digital di masa modern dalam bidang arsitektur perlahan-lahan mulai menggeser cara-cara tradisional dalam proses perencanaan dan perancangan arsitektur. Tulisan ini bertujuan untuk mengkaji peran teknologi digital dengan menggunakan Artificial Intellegence dalam proses perencanaan dan perancangan arsitektur di masa kini sehingga menjadikan proses desain menjadi lebih cepat, mudah, dan praktis serta memberikan pilihan-pilihan yang lebih beragam bagi arsitek untuk menentukan keputusan desain. Metode yang digunakan dalam penulisan ini adalah dengan menganalisa literature dari beberapa penelitian terkait tentang teori perencanaan dan perancangan serta serta Artificial Intellegence. Dengan perkembangan yang cukup eksponensial, penggunaan Artificial Intelligence dalam bidang arsitektur dimungkinkan bukan hanya sebagai alat perpanjangan tangan dari arsitek tetapi juga perpanjangan otak yang dapat mengeksplorasi desain dengan lebih banyak pilihan dalam waktu yang lebih cepat secara efektif dan efisien.

show abstract

“…18 Language-based , which may use various types of neural networks with the aid of Natural Language Processing (NLP) to generate floor plans through linguistic descriptions, 19,20 and last but not least, Pixel-based , which uses the color of pixels to label the room types besides also being able to provide information like shape, orientation or area, these are commonly trained with GANs, 9,21,22 even though, there are methods which does not receive any kind of generative constraint besides the train data itself. 23,24 Thus, in summary, our overview of deep generative methods for floor plans is taxonomized as graph-based, language-based, and pixel-based.…”

Section: Introductionmentioning

confidence: 99%

Generating floor plans with deep learning: A cross-validation assessment over different dataset sizes

Rodrigues

Duarte

2022

International Journal of Architectural Computing

View full text Add to dashboard Cite

The advent of deep learning has enabled a series of opportunities; one of them is the ability to tackle subjective factors on the floor plan design and make predictions though spatial semantic maps. Nonetheless, the amount available of data grows exponentially on a daily basis, in this sense, this research seeks to investigate deep generative methods of floor plan design and its relationship between data volume, with training time, quality and diversity in the outputs; in other words, what is the amount of data required to rapidly train models that return optimal results. In our research, we used a variation of the Conditional Generative Adversarial Network algorithm, that is, Pix2pix, and a dataset of approximately 80 thousand images to train 10 models and evaluate their performance through a series of computational metrics. The results show that the potential of this data-driven method depends not only on the diversity of the training set but also on the linearity of the distribution; therefore, high-dimensional datasets did not achieve good results. It is also concluded that models trained on small sets of data (800 images) may return excellent results if given the correct training instructions (Hyperparameters), but the best baseline to this generative task is in the mid-term, using around 20 to 30 thousand images with a linear distribution. Finally, it is presented standard guidelines for dataset design, and the impact of data curation along the entire process.

show abstract

GAN as a generative architectural plan layout tool: A case study for training DCGAN with Palladian Plans and evaluation of DCGAN outputs

Cited by 11 publications

References 0 publications

The Role of AI Design Assistance on the Architectural Design Process: An Empirical Research with Novice Designers

The Role of AI Design Assistance on the Architectural Design Process: An Empirical Research with Novice Designers

Peran Artifical Intelligence dalam Tahap Perencanaan dan Perancangan Desain Arsitektur

Generating floor plans with deep learning: A cross-validation assessment over different dataset sizes

Contact Info

Product

Resources

About