returning the centroid of each cell in the sample dict (microsoft#1240)

* returning the centroid of each cell in the sample dict

* updating usavars test to reflect centroid location as part of each sample

* returning the centroid of each cell in the sample dict

* updating usavars test to reflect centroid location as part of each sample

* separating lat and lon in each sample

* separating lat and lon in each sample

tests/datasets/test_usavars.py

@@ -90,9 +90,11 @@ def test_getitem(self, dataset: USAVars) -> None:
         assert isinstance(x, dict)
         assert isinstance(x["image"], torch.Tensor)
         assert x["image"].ndim == 3
-        assert len(x.keys()) == 2  # image, labels
+        assert len(x.keys()) == 4  # image, labels, centroid_lat, centroid_lon
         assert x["image"].shape[0] == 4  # R, G, B, Inf
         assert len(dataset.labels) == len(x["labels"])
+        assert len(x["centroid_lat"]) == 1
+        assert len(x["centroid_lon"]) == 1
 
     def test_len(self, dataset: USAVars) -> None:
         if dataset.split == "train":

torchgeo/datasets/usavars.py

@@ -166,6 +166,8 @@ def __getitem__(self, index: int) -> dict[str, Tensor]:
                 [self.label_dfs[lab].loc[id_][lab] for lab in self.labels]
             ),
             "image": self._load_image(os.path.join(self.root, "uar", tif_file)),
+            "centroid_lat": Tensor([self.label_dfs[self.labels[0]].loc[id_]["lat"]]),
+            "centroid_lon": Tensor([self.label_dfs[self.labels[0]].loc[id_]["lon"]]),
         }
 
         if self.transforms is not None: