Our Housing-Design README

This project presents a comprehensive solution to the housing design challenge, rooted in collaboration between architects and advanced technology. Through the use of Heterogeneous Layout Graphs and an effective pipeline, we aim to transform the housing design process, making it more efficient, accessible, and aligned with user requirements while preserving the critical role of architects in the process. We augment the Deep Generative Model of Graphs (DGMG*) from [Li et al., 2018] to create heterogeneous graphs. Hetero-HouseGAN++ from [Nauata et al., 2021] is implemented for the final generation of architecturalquality floor plans.

Steps to install our project (Tested on Ubuntu 20.04):

1. Clone repository:

cd ~
git clone https://github.com/jlhofland/housing-design.git

2. Install conda environment

cd housing-design/
conda env create -f environment.yml
y
conda activate house
pip install matplotlib
pip install scikit-image
pip install opencv-python
pip install svgwrite
pip install webcolors

3. Install our python package

cd housingpipeline
pip install .

Steps to run housing design pipeline

At this point, the user will need to edit files that contain hard-coded file paths to redirect to their install directory. This includes:

• Line 95 in /housing-design/housingpipeline/housingpipeline/floor_plan_pipeline/main.py
• Line 38 in /housing-design/housingpipeline/housingpipeline/floor_plan_pipeline/input_to_graph.py

Then, the user may run the following:

cd ~/housing-design/housingpipeline/housingpipeline/floor_plan_pipeline
python main.py