Skip to content

Latest commit

 

History

History
63 lines (39 loc) · 3.45 KB

README.md

File metadata and controls

63 lines (39 loc) · 3.45 KB
Raw

ObjFormer

Multimodal Change Detection on Paired Map Data and Optical Imagery via Object-Guided Transformer

Hongruixuan Chen1, Cuiling Lan2, Jian Song1,3, Clifford Broni-Bediako3, Junshi Xia3, Naoto Yokoya1,3 *

TGRS paper arXiv paper

1 The University of Tokyo, 2 Microsoft Research Asia, 3 RIKEN AIP, * Corresponding author

Overview | Get Started | Taken Away | Common Issues | Others | 简体中文版

🛎️Updates

  • Notice: ObjFormer has been accepted by IEEE TGRS! We will upload the dataset and code soon. We'd appreciate it if you could give this repo a ⭐️star⭐️ and stay tuned!!
  • July 01st, 2024: We have uploaded OpenMapCD dataset. You are welcome to download and use it!

🔭Overview

  • OpenMapCD is the first benchmark dataset for multimodal change detecton tasks on optical remote sensing imagery and map data, with 1,287 samples from 40 regions across six continents, supoorting both binary and semantic change detection.

  • ObjFormer serves as a robust and efficient benchmark for the proposed multimodal change detection tasks by combining OBIA techniques with self-attention mechanisms.

🗝️Let's Get Started!

A. Installation

Under construction...

📜Reference

If this code or dataset contributes to your research, please kindly consider citing our paper and give this repo ⭐️ :)

@ARTICLE{Chen2024ObjFormer,
  author={Chen, Hongruixuan and Lan, Cuiling and Song, Jian and Broni-Bediako, Clifford and Xia, Junshi and Yokoya, Naoto},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={ObjFormer: Learning Land-Cover Changes From Paired OSM Data and Optical High-Resolution Imagery via Object-Guided Transformer}, 
  year={2024},
  volume={62},
  number={},
  pages={1-22},
  doi={10.1109/TGRS.2024.3410389}
}

🔗Other links

If you are interested in land-cover mapping and domain adaptation in remote sensing using synthetic datasets, you can also follow our two datasets below.

  • OpenEarthMap dataset: a benchmark dataset for global sub-meter level land cover mapping.

  • SyntheWorld dataset: a large-scale synthetic remote sensing datasets for land cover mapping and building change detection.

🙋Q & A

For any questions, please contact us.