Add xView2 Dataset

docs/api/datasets.rst

@@ -182,6 +182,12 @@ UC Merced
 .. autoclass:: UCMerced
 .. autoclass:: UCMercedDataModule
 
+xView2
+^^^^^^
+
+.. autoclass:: XView2
+.. autoclass:: XView2DataModule
+
 ZueriCrop
 ^^^^^^^^^
 

tests/datasets/test_xview2.py

@@ -0,0 +1,117 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+# Licensed under the MIT License.
+
+import os
+import shutil
+from pathlib import Path
+from typing import Generator
+
+import pytest
+import torch
+import torch.nn as nn
+from _pytest.fixtures import SubRequest
+from _pytest.monkeypatch import MonkeyPatch
+
+from torchgeo.datasets import XView2, XView2DataModule
+
+
+class TestXView2:
+    @pytest.fixture(params=["train", "test"])
+    def dataset(
+        self, monkeypatch: Generator[MonkeyPatch, None, None], request: SubRequest
+    ) -> XView2:
+        monkeypatch.setattr(  # type: ignore[attr-defined]
+            XView2,
+            "metadata",
+            {
+                "train": {
+                    "filename": "train_images_labels_targets.tar.gz",
+                    "md5": "373e61d55c1b294aa76b94dbbd81332b",
+                    "directory": "train",
+                },
+                "test": {
+                    "filename": "test_images_labels_targets.tar.gz",
+                    "md5": "bc6de81c956a3bada38b5b4e246266a1",
+                    "directory": "test",
+                },
+            },
+        )
+        root = os.path.join("tests", "data", "xview2")
+        split = request.param
+        transforms = nn.Identity()  # type: ignore[attr-defined]
+        return XView2(root, split, transforms, checksum=True)
+
+    def test_getitem(self, dataset: XView2) -> None:
+        x = dataset[0]
+        assert isinstance(x, dict)
+        assert isinstance(x["image"], torch.Tensor)
+        assert isinstance(x["mask"], torch.Tensor)
+
+    def test_len(self, dataset: XView2) -> None:
+        assert len(dataset) == 2
+
+    def test_extract(self, tmp_path: Path) -> None:
+        shutil.copyfile(
+            os.path.join(
+                "tests", "data", "xview2", "train_images_labels_targets.tar.gz"
+            ),
+            os.path.join(tmp_path, "train_images_labels_targets.tar.gz"),
+        )
+        shutil.copyfile(
+            os.path.join(
+                "tests", "data", "xview2", "test_images_labels_targets.tar.gz"
+            ),
+            os.path.join(tmp_path, "test_images_labels_targets.tar.gz"),
+        )
+        XView2(root=str(tmp_path))
+
+    def test_corrupted(self, tmp_path: Path) -> None:
+        with open(
+            os.path.join(tmp_path, "train_images_labels_targets.tar.gz"), "w"
+        ) as f:
+            f.write("bad")
+        with open(
+            os.path.join(tmp_path, "test_images_labels_targets.tar.gz"), "w"
+        ) as f:
+            f.write("bad")
+        with pytest.raises(RuntimeError, match="Dataset found, but corrupted."):
+            XView2(root=str(tmp_path), checksum=True)
+
+    def test_invalid_split(self) -> None:
+        with pytest.raises(AssertionError):
+            XView2(split="foo")
+
+    def test_not_downloaded(self, tmp_path: Path) -> None:
+        with pytest.raises(RuntimeError, match="Dataset not found in `root` directory"):
+            XView2(str(tmp_path))
+
+    def test_plot(self, dataset: XView2) -> None:
+        x = dataset[0].copy()
+        XView2.plot(x, suptitle="Test")
+        XView2.plot(x, show_titles=False)
+        x["prediction"] = x["mask"][0].clone()
+        XView2.plot(x)
+
+
+class TestXView2DataModule:
+    @pytest.fixture(scope="class", params=[0.0, 0.5])
+    def datamodule(self, request: SubRequest) -> XView2DataModule:
+        root = os.path.join("tests", "data", "xview2")
+        batch_size = 1
+        num_workers = 0
+        val_split_size = request.param
+        dm = XView2DataModule(
+            root, batch_size, num_workers, val_split_pct=val_split_size
+        )
+        dm.prepare_data()
+        dm.setup()
+        return dm
+
+    def test_train_dataloader(self, datamodule: XView2DataModule) -> None:
+        next(iter(datamodule.train_dataloader()))
+
+    def test_val_dataloader(self, datamodule: XView2DataModule) -> None:
+        next(iter(datamodule.val_dataloader()))
+
+    def test_test_dataloader(self, datamodule: XView2DataModule) -> None:
+        next(iter(datamodule.test_dataloader()))

torchgeo/datasets/__init__.py

@@ -62,6 +62,7 @@
 from .spacenet import SpaceNet, SpaceNet1, SpaceNet2, SpaceNet4, SpaceNet7
 from .ucmerced import UCMerced, UCMercedDataModule
 from .utils import BoundingBox, collate_dict
+from .xview import XView2, XView2DataModule
 from .zuericrop import ZueriCrop
 
 __all__ = (
@@ -130,6 +131,8 @@
     "UCMerced",
     "UCMercedDataModule",
     "VHR10",
+    "XView2",
+    "XView2DataModule",
     "ZueriCrop",
     # Base classes
     "GeoDataset",

torchgeo/datasets/utils.py

@@ -20,6 +20,7 @@
 from torch import Tensor
 from torch.utils.data import Dataset, Subset, random_split
 from torchvision.datasets.utils import check_integrity, download_url
+from torchvision.utils import draw_segmentation_masks
 
 __all__ = (
     "check_integrity",
@@ -33,9 +34,13 @@
     "rasterio_loader",
     "dataset_split",
     "sort_sentinel2_bands",
+    "draw_semantic_segmentation_masks",
 )
 
 
+ColorMap = Union[List[Union[str, Tuple[int, int, int]]], str, Tuple[int, int, int]]
+
+
 class _rarfile:
     class RarFile:
         def __init__(self, *args: Any, **kwargs: Any) -> None:
@@ -431,3 +436,26 @@ def sort_sentinel2_bands(x: str) -> str:
     if x == "B8A":
         x = "B08A"
     return x
+
+
+def draw_semantic_segmentation_masks(
+    image: Tensor, mask: Tensor, alpha: float = 0.5, colors: Optional[ColorMap] = None
+) -> np.ndarray:  # type: ignore[type-arg]
+    """Overlay a semantic segmentation mask onto an image.
+
+    Args:
+        image: tensor of shape (3, h, w)
+        mask: tensor of shape (h, w) with pixel values representing the classes
+        alpha: alpha blend factor
+        colors: (Optional) list of RGB int tuples, or color strings e.g. red, #FF00FF
+    Returns:
+        a list of the subset datasets. Either [train, val] or [train, val, test]
+    """
+    classes = torch.unique(mask)  # type: ignore[attr-defined]
+    classes = classes[1:]
+    class_masks = mask == classes[:, None, None]
+    img = draw_segmentation_masks(
+        image=image, masks=class_masks, alpha=alpha, colors=colors
+    )
+    img = img.permute((1, 2, 0)).numpy()
+    return img  # type: ignore[no-any-return]