Information and code about the IGARSS 2022 publication

WUP-CD: Towards 2.5D Data for Deep Learning Building Change Detection

Download the Datasets

WUPpertal building Change Detection dataset (WUP-CD) The datasaets are kindly hosted by zenodo.org.

Medium Resolution

• 0.5m Ground Sampling Distance

WUP-CD_MR

High Resolution

• 0.2m Ground Sampling Distance

WUP-CD_HR

Train a model

Use the script train.py to train any of the DDSM-based models presented in the paper.

Test a model

Use the script test.py to test a trained and saved model.

Citation

If you find this work or the data itself useful, please cite us.

@INPROCEEDINGS{9883863,
  author={Bauer, Adrian and Oberbossel, Jens and Sander, Stefan and Kummert, Anton},
  booktitle={IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium}, 
  title={WUP-CD: Towards 2.5D Data for Deep Learning Building Change Detection}, 
  year={2022},
  volume={},
  number={},
  pages={219-222},
  keywords={Deep learning;Analytical models;Image resolution;Buildings;Data acquisition;Feature extraction;Data models;Change Detection (CD);Remote Sensing Dataset;Deep Learning;Digital Surface Model;2.5D Data},
  doi={10.1109/IGARSS46834.2022.9883863}}

License

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.