This is an official implementation of STAR and ChangeStar in our ICCV 2021 paper Change is Everywhere: Single-Temporal Supervised Object Change Detection for High Spatial Resolution Remote Sensing Imagery.
We hope that STAR will serve as a solid baseline and help ease future research in weakly-supervised object change detection.
pip install ever-beta==0.2.3
Download xView2 dataset (training set and tier3 set) and LEVIR-CD dataset.
Create soft link
ln -s </path/to/xView2> ./xview2
ln -s </path/to/LEVIR-CD> ./LEVIR-CD
bash ./scripts/trainxView2/r50_farseg_changemixin_symmetry.sh
bash ./scripts/bisup_levircd/r50_farseg_changemixin.sh
If you use STAR or ChangeStar (FarSeg) in your research, please cite the following paper:
@inproceedings{zheng2021change,
title={Change is Everywhere: Single-Temporal Supervised Object Change Detection in Remote Sensing Imagery},
author={Zheng, Zhuo and Ma, Ailong and Zhang, Liangpei and Zhong, Yanfei},
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
pages={15193--15202},
year={2021}
}
@inproceedings{zheng2020foreground,
title={Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery},
author={Zheng, Zhuo and Zhong, Yanfei and Wang, Junjue and Ma, Ailong},
booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
pages={4096--4105},
year={2020}
}
This code is released under the Apache License 2.0.
Copyright (c) Zhuo Zheng. All rights reserved.