PanCollection

April 7, 2023 · View on GitHub

Pansharpening Datasets from WorldView 2, WorldView 3, QuickBird, Gaofen 2 sensors

[Chinese Webpage]
Recommendation: Use the code-toolbox [DLPan-Toolbox] + the dataset [PanCollection] for fair training and testing!
Also, a dataset [HyperPanCollection] for another similar task, i.e., hyperspectral pansharpening!
Latest Update (Dec. 11, 2022): we updated full-resolution test examples that contain more different imgae scenes.
Latest Update (Mar. 20, 2023): one testing example in reduce-resolution format for WV3 sensor is not consistent with the one in full-resolution format, we have fixed it.

Download by `Google Drive`

(1) The training and testing datasets for `WorldView 3`:

WorldView 3 Dataset	Link	Size
Training Dataset	[download link]	5.76GB
Testing Dataset (ReducedData, H5 Format)	[download link]	20 examples
Testing Dataset (FullData, H5 Format)	[download link]	20 examples
Testing Dataset (ReducedData, mat Format)	[download link]	20 examples
Testing Dataset (FullData, mat Format)	[download link]	20 examples

Note: H5 files have same data with mat files (but with different formats) which can be used for single image test

(2) The training and testing datasets for `QuickBird`:

QuickBird Dataset	Link	Size
Training Dataset	[download link]	5.37GB
Testing Dataset (ReducedData, H5 Format)	[download link]	20 examples
Testing Dataset (FullData, H5 Format)	[download link]	20 examples
Testing Dataset (ReducedData, mat Format)	[download link]	20 examples
Testing Dataset (FullData, mat Format)	[download link]	20 examples

(3) The training and testing datasets for `Gaofen 2`:

Gaofen 2 Dataset	Link	Size
Training Dataset	[download link]	6.21GB
Testing Dataset (ReducedData, H5 Format)	[download link]	20 examples
Testing Dataset (FullData, H5 Format)	[download link]	20 examples
Testing Dataset (ReducedData, mat Format)	[download link]	20 examples
Testing Dataset (FullData, mat Format)	[download link]	20 examples

(4) The testing datasets for `WorldView 2`:

WorldView 2 Dataset	Link	Size
Testing Dataset (ReducedData, H5 Format)	[download link]	20 examples
Testing Dataset (FullData, H5 Format)	[download link]	20 examples
Testing Dataset (ReducedData, mat Format)	[download link]	20 examples
Testing Dataset (FullData, mat Format)	[download link]	20 examples

Note: This data is only used for the test of network generalization, thus no training dataset!

Download by `Baidu Cloud`

Use [Baidu Cloud] to download these datasets.

(1) The training and testing datasets for `WorldView 3`:

WorldView 3 Dataset	Link	Size
Training Dataset	[download link]	5.76GB
Testing Dataset (ReducedData, H5 Format)	[download link]	20 examples
Testing Dataset (FullData, H5 Format)	[download link]	20 examples
Testing Dataset (ReducedData, mat Format)	[download link]	20 examples
Testing Dataset (FullData, mat Format)	[download link]	20 examples

(2) The training and testing datasets for `QuickBird`:

QuickBird Dataset	Link	Size
Training Dataset	[download link]	5.37GB
Testing Dataset (ReducedData, H5 Format)	[download link]	20 examples
Testing Dataset (FullData, H5 Format)	[download link]	20 examples
Testing Dataset (ReducedData, mat Format)	[download link]	20 examples
Testing Dataset (FullData, mat Format)	[download link]	20 examples

(3) The training and testing datasets for `Gaofen 2`:

Gaofen 2 Dataset	Link	Size
Training Dataset	[download link]	6.21GB
Testing Dataset (ReducedData, H5 Format)	[download link]	20 examples
Testing Dataset (FullData, H5 Format)	[download link]	20 examples
Testing Dataset (ReducedData, mat Format)	[download link]	20 examples
Testing Dataset (FullData, mat Format)	[download link]	20 examples

(4) The testing datasets for `WorldView 2`:

WorldView 2 Dataset	Link	Size
Testing Dataset (ReducedData, H5 Format)	[download link]	20 examples
Testing Dataset (FullData, H5 Format)	[download link]	20 examples
Testing Dataset (ReducedData, mat Format)	[download link]	20 examples
Testing Dataset (FullData, mat Format)	[download link]	20 examples

Note: This data is only used for the test of network generalization, thus no training dataset!

Reference

More details about the similation procedure of datasets, you may check the following two papers:

@ARTICLE{dengjig2022,
	author={邓良剑，冉燃，吴潇，张添敬},
	journal={中国图象图形学报},
	title={遥感图像全色锐化的卷积神经网络方法研究进展},
 	year={2022},
  	volume={},
  	number={9},
  	pages={},
  	doi={10.11834/jig.220540}
   }

and

@ARTICLE{deng2022vivone,
	author={L. -J. Deng, G. Vivone, M. E. Paoletti, G. Scarpa, J. He, Y. Zhang, J. Chanussot, and A. Plaza},
	journal={IEEE Geoscience and Remote Sensing Magazine}, 
	title={Machine Learning in Pansharpening: A Benchmark, from Shallow to Deep Networks}, 
	year={2022},
	volume={10},
	number={3},
	pages={279-315},
	doi={10.1109/MGRS.2022.3187652}
   }

Contact:

We are glad to hear from you. If you have any questions, please feel free to contact wxwsx1997@qq.com.