DMGM

Code and dataset for ICRA2023 paper "Deep Masked Graph Matching for Correspondence Identification in Collaborative Perception"

Overview

The overview of our approach is shown as follows:

Dataset

1. To test our approach, we upload 2 processed data instances, which can be found in testdata/data folder.

2. Our test dataset can be found here

3. The full dataset (scenario 2) used in this paper can be found here

4. Our extended dataset (scenarios 1,3,4,5) can be found here

Requirements

Our work is based on Pytorch and Pytorch-Geometric, we recommend using conda environment to run our code.

---

conda create --name DMGM python=3.9
conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3 -c pytorch
conda install pyg -c pyg
pip install -r requirements.txt

---

Test

We provide pre-trained model in checkpoint folder.

If you just want to evaluate the test dataset, then please move the test_dataset to testdata/data folder and run the following command.

---

python run test_rural_ours.py

---

If you use our full_dataset, then you may change the path in "test_rural_ours.py" as follows:

---

in_path_s = os.path.join('$fulldataset/proc_s')
in_path_t = os.path.join('$fulldataset/proc_t')

---

and then run the "test_rural_ours.py".

Citation

If you use DMGM in a scientific publication, we would appreciate using the following citation:

---

@inproceedings{gao2023deep, 
author    = {Gao, Peng and Zhu, Qingzhao and Lu, Hongsheng and Gan, Chuang and Zhang, Hao}, 
title     = {{Deep Masked Graph Matching for Correspondence Identification in Collaborative Perception}}, 
booktitle = {International Conference on Robotics and Automation (ICRA)}, 
year      = {2023},
url      = {https://arxiv.org/abs/2303.07555} 

}