This repository contains the code of GNN-DM, a graph neural network for gas distribution mapping (GDM). GDM describes the process of mapping the spatial and temporal distributions of gases in a given area.
The synthetic gas distribution dataset is based on the dataset that was previously made available in the repository of Super-Resolution for Gas Distribution Mapping. In the linked repo, training and validation datasets are available through Git LFS (see data/30x25.zip). Unzip the files to this directory: data/30x25/raw.
If you find this code useful, please cite our paper:
@inproceedings{winkler2024rabignn,
title={Radius-Based, Bi-Directional Graph Neural Networks for Gas Distribution Mapping (RABI-GNN)},
author={Winkler, Nicolas P and Neumann, Patrick P and Schaffernicht, Erik and Lilienthal, Achim J},
booktitle={2024 IEEE International Symposium on Olfaction and Electronic Nose (ISOEN)},
pages={1--3},
year={2024},
organization={IEEE}
}