Add CanadianBuildingFootprints dataset

.gitattributes

@@ -1,2 +1,2 @@
 # Do not change line endings on test data, it will change the MD5
-/tests/data/** binary
+/tests/data/*/** binary

.github/workflows/tests.yaml

@@ -62,7 +62,7 @@ jobs:
       if: ${{ runner.os == 'Windows' }}
     - name: Install conda dependencies (Windows)
       run: |
-        conda install h5py 'rasterio>=1.0'
+        conda install fiona h5py 'rasterio>=1.0'
         conda list
         conda info
       if: ${{ runner.os == 'Windows' }}

docs/datasets.rst

@@ -10,6 +10,11 @@ Geospatial Datasets
 
 :class:`GeoDataset` is designed for datasets that contain geospatial information, like latitude, longitude, coordinate system, and projection. Datasets containing this kind of information can be combined using :class:`ZipDataset`.
 
+Canadian Building Footprints
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+.. autoclass:: CanadianBuildingFootprints
+
 Chesapeake Bay High-Resolution Land Cover Project
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 

environment.yml

@@ -3,6 +3,7 @@ channels:
   - conda-forge
 dependencies:
   - cudatoolkit
+  - fiona
   - h5py
   - numpy
   - pip
@@ -17,6 +18,7 @@ dependencies:
     - black[colorama]>=21b
     - flake8
     - isort[colors]>=4.3.5
+    - jupyterlab
     - mypy>=0.900
     - omegaconf
     - opencv-python

requirements.txt

@@ -1,8 +1,10 @@
 affine
 black[colorama]>=21b
+fiona
 flake8
 h5py
 isort[colors]>=4.3.5
+jupyterlab
 matplotlib
 mypy>=0.900
 numpy

setup.cfg

@@ -26,6 +26,7 @@ setup_requires =
     setuptools>=42
 install_requires =
     affine
+    fiona
     matplotlib
     numpy
     pillow
@@ -50,6 +51,7 @@ datasets =
 
 # Optional developer requirements
 docs =
+    jupyterlab
     sphinx
     pydocstyle[toml]>=6.1
     pytorch-sphinx-theme

spack.yaml

@@ -5,9 +5,11 @@ spack:
   - "python@3.7:+bz2"
   - py-affine
   - "py-black@21:+colorama"
+  - py-fiona
   - py-flake8
   - py-h5py
   - "py-isort@4.3.5:+colors"
+  - py-jupyterlab
   - py-matplotlib
   - "py-mypy@0.900:"
   - py-numpy

tests/data/README.md

@@ -1,9 +1,11 @@
-This directory contains fake data used to test torchgeo. Depending on the type of dataset, fake data can be created in one of two ways:
+This directory contains fake data used to test torchgeo. Depending on the type of dataset, fake data can be created in multiple ways:
 
 ## GeoDataset
 
 GeoDataset data can be created like so. We first open an existing data example and use it to copy the driver/CRS/transform to the fake data.
 
+### Raster data
+
 ```python
 import os
 
@@ -25,7 +27,21 @@ cmap = src.colormap(1)
 dst.write_colormap(1, cmap)
 ```
 
-If the dataset expects multiple files, you can simply copy and rename the file you created.
+### Vector data
+
+```python
+import os
+
+import fiona
+
+ROOT = "/mnt/blobfuse/adam-scratch/cbf"
+FILENAME = "Ontario.geojson"
+
+src = fiona.open(os.path.join(ROOT, FILENAME))
+dst = fiona.open(FILENAME, "w", **src.meta)
+rec = next(iter(src))
+dst.write(rec)
+```
 
 ## VisionDataset
 

tests/datasets/test_cbf.py

@@ -0,0 +1,92 @@
+import os
+import shutil
+from pathlib import Path
+from typing import Generator
+
+import matplotlib.pyplot as plt
+import pytest
+import torch
+from _pytest.fixtures import SubRequest
+from pytest import MonkeyPatch
+from rasterio.crs import CRS
+
+import torchgeo.datasets.utils
+from torchgeo.datasets import BoundingBox, CanadianBuildingFootprints, ZipDataset
+from torchgeo.transforms import Identity
+
+
+def download_url(url: str, root: str, *args: str) -> None:
+    shutil.copy(url, root)
+
+
+class TestCanadianBuildingFootprints:
+    @pytest.fixture
+    def dataset(
+        self,
+        monkeypatch: Generator[MonkeyPatch, None, None],
+        tmp_path: Path,
+        request: SubRequest,
+    ) -> CanadianBuildingFootprints:
+        monkeypatch.setattr(  # type: ignore[attr-defined]
+            torchgeo.datasets.utils, "download_url", download_url
+        )
+        md5s = [
+            "aef9a3deb3297f225d6cdb221cb48527",
+            "2b7872c4121116fda8f96490daf89d29",
+            "c71ded923e22569b62b00da2d2a61076",
+            "75a8f652531790c3c3aefc0655400d6d",
+            "89ff9c6257efa99365a8b709dde9579b",
+            "d4d6a36ed834df5cbf5254effca78a4d",
+            "cce85f6183427e3034704cf35919c985",
+            "0149c7ec5101c0309c79b7e695dcb394",
+            "b05216155725f48937804371b945f8ae",
+            "72d0e6d7196345ca520c825697cc4947",
+            "77e1c6c71ff0efbdd221b7e7d4a5f2df",
+            "86e32374f068c7bbb76aa81af0736733",
+            "5e453a3426b0bb986b2837b85e8b8850",
+        ]
+        monkeypatch.setattr(  # type: ignore[attr-defined]
+            CanadianBuildingFootprints, "md5s", md5s
+        )
+        url = os.path.join("tests", "data", "cbf") + os.sep
+        monkeypatch.setattr(  # type: ignore[attr-defined]
+            CanadianBuildingFootprints, "url", url
+        )
+        monkeypatch.setattr(  # type: ignore[attr-defined]
+            plt, "show", lambda *args: None
+        )
+        (tmp_path / "cbf").mkdir()
+        root = str(tmp_path)
+        transforms = Identity()
+        return CanadianBuildingFootprints(
+            root, transforms=transforms, download=True, checksum=True
+        )
+
+    def test_getitem(self, dataset: CanadianBuildingFootprints) -> None:
+        x = dataset[dataset.bounds]
+        assert isinstance(x, dict)
+        assert isinstance(x["crs"], CRS)
+        assert isinstance(x["masks"], torch.Tensor)
+
+    def test_add(self, dataset: CanadianBuildingFootprints) -> None:
+        ds = dataset + dataset
+        assert isinstance(ds, ZipDataset)
+
+    def test_already_downloaded(self, dataset: CanadianBuildingFootprints) -> None:
+        CanadianBuildingFootprints(root=dataset.root, download=True)
+
+    def test_plot(self, dataset: CanadianBuildingFootprints) -> None:
+        query = dataset.bounds
+        x = dataset[query]
+        dataset.plot(x["masks"])
+
+    def test_not_downloaded(self, tmp_path: Path) -> None:
+        with pytest.raises(RuntimeError, match="Dataset not found or corrupted."):
+            CanadianBuildingFootprints(str(tmp_path))
+
+    def test_invalid_query(self, dataset: CanadianBuildingFootprints) -> None:
+        query = BoundingBox(0, 0, 0, 0, 0, 0)
+        with pytest.raises(
+            IndexError, match="query: .* is not within bounds of the index:"
+        ):
+            dataset[query]

torchgeo/datasets/__init__.py

@@ -1,6 +1,7 @@
 """TorchGeo datasets."""
 
 from .benin_cashews import BeninSmallHolderCashews
+from .cbf import CanadianBuildingFootprints
 from .cdl import CDL
 from .chesapeake import (
     Chesapeake,
@@ -42,6 +43,7 @@
 __all__ = (
     "BeninSmallHolderCashews",
     "BoundingBox",
+    "CanadianBuildingFootprints",
     "CDL",
     "collate_dict",
     "Chesapeake",