HouseDiffusion: Vector Floorplan Generation via a Diffusion Model with Discrete and Continuous Denoising
1. Clone repo and install the requirements:
My implementation is based on the public implementation of HouseDiffusion.
git clone https://github.com/sakmalh/house_diffusion
cd house_diffusion
pip install -r requirements.txt
2. Initiate the Backend
uvicorn app:app --host 0.0.0.0
- The pretrained model will be downloaded automatically from the Google Drive. pretrained
1. Request and Response Types
- To test the endpoint you could simply access http://0.0.0.0:8080/generate
{
"nodes": [
{
"id": "string", # Nodes unique id
"room_type": "string", # Room Type in ['Living Room', 'Kitchen', 'Bedroom', 'Bathroom', 'Balcony', 'Entrance', 'Dining Room', 'Study Room', 'Storage', 'Front Door', 'Unknown', 'Interior Door']
"corners": "string" # Number of Corners
}
],
"edges": [
{
"id": "string", # Edge Unique id
"source": "string", # First Nodes id
"target": "string" # Second Nodes id
}
],
"metrics": true # If True provides the length and width of rooms in pixels
}
- DockerFile is provided for hosting.
docker build -t {docker-repo}/backend .
@article{shabani2022housediffusion,
title={HouseDiffusion: Vector Floorplan Generation via a Diffusion Model with Discrete and Continuous Denoising},
author={Shabani, Mohammad Amin and Hosseini, Sepidehsadat and Furukawa, Yasutaka},
journal={arXiv preprint arXiv:2211.13287},
year={2022}
}