HR-SARBuD (High Resolution SAR Building Dataset) is created based on SpaceNet6 for deep learning(Details are available at https://spacenet.ai/sn6-challenge/). The dataset was originally acquired from Capella SAR satellite data (0.5m). It contains building patches of SAR image and corresponding label images. This repository was derived from co-polarization bands (HH and VV) of the SAR-Intensity folder. There are about 22000 SAR image patches of building area and corresponding 22000 labels (Where 0 means background and 1 means building) in current version. The size of the images were set to 512×512. The dataset is compressed into zip format, with a total of 3 files.
Our dataset is so large that you cannot download it direcly, You can download it in two ways.
You can download the dataset and checkpoint weight from Google Drive
You can download the dataset and checkpoint weight from Baidu Netdisk.
Shermeyer, J., Hogan, D., Brown, J., Etten, A.V., Weir, N., Pacifici, F., Hänsch, R., Bastidas, A., Soenen, S., Bacastow, T.M., & Lewis, R. (2020). SpaceNet 6: Multi-Sensor All Weather Mapping Dataset. 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 768-777.