Publication
Generation of Hospital Emergency Department Layouts Based on Generative Adversarial Networks
Generating hospital emergency department layouts with hierarchical design and generative adversarial networks.
Cite this work View paper
Citation
Zhao, C. W., Yang, J., & Li, J. (2021). Generation of hospital emergency department layouts based on generative adversarial networks. Journal of Building Engineering, 43, 102539.
About this work
This study compares DCGAN, pix2pix, and CycleGAN for generating hospital emergency department layouts. The models are trained on 120 annotated sets of department drawings and evaluated from architectural and algorithmic perspectives.
The comparison examines layout flexibility, image quality, and dataset requirements. CycleGAN gives useful layouts with fewer constraints on the training data, while pix2pix produces the highest image quality.