We build a challenging PolSAR terrain segmentation data set called AIR-PolSAR-Seg with higher data scale and scene complexity. The data set is open to the community to provide support for research in the field of PolSAR terrain segmentation.

The presented AIR-PolSAR-Seg contain 500 full-polarization SAR images with a size of 512x512 and each image has 4 kinds of polarization modes, including the HH,HV,VH,and VV. Therefore, the total number is 2000. These full-polarization SAR images are provided by the Gaofen-3 satellite at quad-polarized strip I (QPSI) mode. The spatial resolution is 8m, and it is annotated with respect to 6 categories at pixel level.

This data set can be download from：

BaiduNetdisk:https://pan.baidu.com/s/1PWsEwtdNKCdghthBnhb8XQ password：ni88 GooleDrive:https://drive.google.com/drive/folders/1sGrmLmknhQ28nvasKbCqKvVj6rY2fCqL?usp=sharing

The link of code can be obtained from:

https://github.com/open-mmlab/mmsegmentation

https://github.com/rmislam/PythonSIFT

https://github.com/jianlong-yuan/HOG-SVM-python

https://github.com/milaan9/Python_Decision_Tree_and_Random_Forest

https://github.com/JeanKossaifi/python-hog

The PolSAR data set, relying on the major scientific and technological projects of the national high-resolution earth observation system, aims to further promote the research of advanced technologies on SAR semantic segmentation. The ownership of the PolSAR data set belongs to the major science and technology project of the national high-resolution earth observation system and the Aerospace Information Research Institute, Chinese Academy of Sciences.