Fix format expected by torchvision utils

requirements.txt

@@ -2,6 +2,7 @@ black
 flake8
 flake8-import-order
 mypy
+PIL
 pycocotools
 rarfile
 torchvision

torchsat/datasets/nwpu.py

@@ -83,7 +83,9 @@ class VHR10(VisionDataset):
         """
         id = self.ids[index]
         image = self._load_image(id)
-        target = self._load_target(id)
+        annot = self._load_target(id)
+
+        target = dict(image_id=id, annotations=annot)
 
         if self.transforms is not None:
             image, target = self.transforms(image, target)