AAG-FP

This project presents a framework for extracting attributed adjacency graphs (AAGs) from floor plan images. paper

Dependencies

Linux or macOS is required
Python ≥ 3.6
detectron2 # need to be installed separately
pytorch # need to be installed separately
Other required packages are summarized in requirements.txt.

Quick start

Download the repository and install the dependencies

git clone https://github.com/JanineCHEN/AAG-FP.git 
cd ~/AAG_FP/
conda create --name AAG_FP python=3.6 # can use either anaconda or virtualenvwrapper to create the virtal environment
conda activate AAG_FP
# detectron2 and pytorch need to be installed separately
pip install -r requirements.txt

Download the checkpoint

For downloading the checkpoints, please refer to ckpt.

Run the framework

This demo example uses the sample floor plan images in FP_sample_images.

you can use your own floor plan images by putting them inside the FP_sample_images folder, images with extension ".jpeg",".jpg" or ".png" are all accepted.

For executing the AAG extractor, please run:

python main.py

Download the data

For downloading the dataset, please refer to dataset.

Citation

If you find the code in our research useful, please consider cite:

@inproceedings{chen_graph_2022,
	title = {Robust attributed adjacency graph extraction using floor plan images},
	volume = {2},
	booktitle = {{POST-CARBON}, {Proceedings} of the 27th {International} {Conference} of the {Association} for {Computer}-{Aided} {Architectural} {Design} {Research} in {Asia}},
	author = {Chen, Jielin and Stouffs, Rudi},
	year = {2022},
	pages = {385--394},
}

Acknowledgments

Part of the code is inspired by CubiCasa/CubiCasa5k and yu45020/Text_Segmentation_Image_Inpainting. The computational work for this project was partially performed on resources of the National Supercomputing Centre, Singapore (https://www.nscc.sg).

EmanuelKuhn/AAG-FP