This is the official repository of [SolarDK: A high-resolution urban solar panel image classification and localization dataset][1] The structure of the dataset is as follows: ``` solardk_dataset_neurips_v2/ ├── bbr │   └── positive ├── gentofte_trainval │   ├── train │   │   ├── mask │   │   ├── negative │   │   └── positive │   └── val │   ├── mask │   ├── negative │   └── positive └── herlev_test └── test ├── mask ├── negative └── positive ``` **Recommended download instructions** 1. Simply `pip install osfclient` 2. Download all the files using: `osf -p aj539 clone <output_directory>` 3. `cd` into downloaded folder 4. Concat & Extract: `cat solardk_dataset_part_* | tar -xvzf -` *Note: You do not need to register as a contributor to download these files* **If you found this dataset useful please consider citing our works:** ``` Khomiakov, M., Radzikowski, J.H., Schmidt, C.A., Sørensen, M.B., Andersen, M., Andersen, M.R. and Frellsen, J., 2022. SolarDK: A high-resolution urban solar panel image classification and localization dataset. arXiv preprint arXiv:2212.01260. ``` [1]: https://arxiv.org/pdf/2212.01260v1.pdf