数据集相关的.md

数据集 Download

多源遥感数据 Spacenet

• 官方链接

• Paper

• 数据集下载
  先安装AWS CLI
  创建一个用于存放数据集的spacenet文件夹，文件夹下打开终端 终端内执行以下命令

  # Training
  aws s3 cp --no-sign-request  s3://spacenet-dataset/spacenet/SN6_buildings/tarballs/SN6_buildings_AOI_11_Rotterdam_train.tar.gz .
  
  # Testing
  aws s3 cp --no-sign-request  s3://spacenet-dataset/spacenet/SN6_buildings/tarballs/SN6_buildings_AOI_11_Rotterdam_test_public.tar.gz .
  

• Dataset structure

   spacenet
    └── SN6_buildings
        │ 
        ├── train/AOI_11_Rotterdam
        │   │
        │   ├── SAR-Intensity	# Tiled geotiffs of 4-Band Quad-Polarized (HH, HV, VH, VV) raster data displaying intensity of backscatter in decibels from Capella’s aerial collect. 
        │   ├── MS      		# Tiled geotiffs of 4-Band Multi-Spectral raster data from Maxar WorldView-2
        │   ├── PAN     		# Tiled geotiffs of Panchromatic raster data from Maxar WorldView-2
        │   ├── PS-MS		# Tiled geotiffs of 4-Band Multi-Spectral raster data pan-sharpened to 0.5m from Maxar WorldView-2
        │   ├── PS-RGB		# Tiled 8-bit color-corrected geotiffs of RGB raster data from Maxar WorldView-2 pan-sharpened to 0.5m
        │   ├── geojson_buildings   # GeoJson labels of building footprints for each tile
        │   └── SummaryData	        # CSV of building footprint locations in pixel coordinates and orientation file indicating the directions from which each SAR image is captured (0 North, 1 South).
        │
        ├── test_public/AOI_11_Rotterdam
        │   │
        │   └── SAR-Intensity	# Tiled geotiffs of 4-Band Quad-Polarized (HH, HV, VH, VV) raster data displaying intensity of backscatter in decibels from Capella’s aerial collect. 
        │
        └── tarballs
            │
            ├── SN6_buildings_AOI_11_Rotterdam_train.tar.gz            # Tarball of all training data (40 GB)
            └── SN6_buildings_AOI_11_Rotterdam_test_public.tar.gz	   # Tarball of all public test data (17 GB)