iSAID_patches

iSAID: A Large-scale Dataset for Instance Segmentation in Aerial Images, CVPR workshops, 2019.

Datasets for Instance Segmentation

1. Data Preparation for Training, Validation and Testing

   1. Please download preprocessed iSAID dataset that contains image segmentation masks. Link

   Make sure that the final dataset must have this structure:

   new_iSAID_patches
   ├── SingleScale
   │   ├── test
   │   │   └── Label.json
   │   │   └── Image
   │   │       ├── P0006_0_0_800_800.png
   │   │       └── ...
   │   │       └── P0009_0_0_800_800.png
   │   ├── train
   │   │   └── Label.json
   │   │   └── Image
   │   │       ├── P0002_0_800_800.png
   │   │       ├── ...
   │   │       └── P0010_0_800_800.png
   │   └── val
   │       └── Label.json
   │       └── Image
   │           ├── P0003_0_0_800_800.png
   │           ├── ...
   │           └── P0004_0_0_800_800.png
   └── MultiScale
       ├── test
       │   └── Label.json
       │   └── Image
       │       ├── P0006_0_0_800_800.png
       │       └── ...
       │       └── P0009_0_0_800_800.png
       ├── train
       │   └── Label.json
       │   └── Image
       │       ├── P0002_0_800_800.png
       │       ├── ...
       │       └── P0010_0_800_800.png
       └── val
           └── Label.json
           └── Image
               ├── P0003_0_0_800_800.png
               ├── ...
               └── P0004_0_0_800_800.png