Add BRIGHT dataset (#2520)

* bright

* bright tests

* bright

* run ruff

* mypy and docs

* ruff on data.py

* ruff on bright

* docs

* ruff

* rm datamodule

* coverage

* request

* Update docs/api/datasets/non_geo_datasets.csv

Co-authored-by: Adam J. Stewart <[email protected]>

---------

Co-authored-by: Adam J. Stewart <[email protected]>

Author: nilsleh

docs/api/datasets.rst

@@ -221,6 +221,11 @@ BioMassters
 
 .. autoclass:: BioMassters
 
+BRIGHT
+^^^^^^
+
+.. autoclass:: BRIGHTDFC2025
+
 CaBuAr
 ^^^^^^
 

docs/api/datasets/non_geo_datasets.csv

@@ -3,6 +3,7 @@ Dataset,Task,Source,License,# Samples,# Classes,Size (px),Resolution (m),Bands
 `Benin Cashew Plantations`_,S,Airbus Pléiades,"CC-BY-4.0",70,6,"1,122x1,186",10,MSI
 `BigEarthNet`_,C,Sentinel-1/2,"CDLA-Permissive-1.0","590,326",19--43,120x120,10,"SAR, MSI"
 `BioMassters`_,R,Sentinel-1/2 and Lidar,"CC-BY-4.0",,,256x256, 10, "SAR, MSI"
+`BRIGHT`_,CD,"MAXAR, NAIP, Capella, Umbra","CC-BY-4.0 AND CC-BY-NC-4.0",3239,4,"0.1-1","RGB,SAR"
 `CaBuAr`_,CD,Sentinel-2,"OpenRAIL",424,2,512x512,20,MSI
 `CaFFe`_,S,"Sentinel-1, TerraSAR-X, TanDEM-X, ENVISAT, ERS-1/2, ALOS PALSAR, and RADARSAT-1","CC-BY-4.0","19092","2 or 4","512x512",6-20,"SAR"
 `ChaBuD`_,CD,Sentinel-2,"OpenRAIL",356,2,512x512,10,MSI

tests/data/bright/data.py

@@ -0,0 +1,117 @@
+#!/usr/bin/env python3
+
+# Copyright (c) Microsoft Corporation. All rights reserved.
+# Licensed under the MIT License.
+
+import hashlib
+import os
+import shutil
+
+import numpy as np
+import rasterio
+
+ROOT = '.'
+DATA_DIR = 'dfc25_track2_trainval'
+
+TRAIN_FILE = 'train_setlevel.txt'
+HOLDOUT_FILE = 'holdout_setlevel.txt'
+VAL_FILE = 'val_setlevel.txt'
+
+TRAIN_IDS = [
+    'bata-explosion_00000049',
+    'bata-explosion_00000014',
+    'bata-explosion_00000047',
+]
+HOLDOUT_IDS = ['turkey-earthquake_00000413']
+VAL_IDS = ['val-disaster_00000001', 'val-disaster_00000002']
+
+SIZE = 32
+
+
+def make_dirs() -> None:
+    paths = [
+        os.path.join(ROOT, DATA_DIR),
+        os.path.join(ROOT, DATA_DIR, 'train', 'pre-event'),
+        os.path.join(ROOT, DATA_DIR, 'train', 'post-event'),
+        os.path.join(ROOT, DATA_DIR, 'train', 'target'),
+        os.path.join(ROOT, DATA_DIR, 'val', 'pre-event'),
+        os.path.join(ROOT, DATA_DIR, 'val', 'post-event'),
+        os.path.join(ROOT, DATA_DIR, 'val', 'target'),
+    ]
+    for p in paths:
+        os.makedirs(p, exist_ok=True)
+
+
+def write_list_file(filename: str, ids: list[str]) -> None:
+    file_path = os.path.join(ROOT, DATA_DIR, filename)
+    with open(file_path, 'w') as f:
+        for sid in ids:
+            f.write(f'{sid}\n')
+
+
+def write_tif(filepath: str, channels: int) -> None:
+    data = np.random.randint(0, 255, (channels, SIZE, SIZE), dtype=np.uint8)
+    # transform = from_origin(0, 0, 1, 1)
+    crs = 'epsg:4326'
+    with rasterio.open(
+        filepath,
+        'w',
+        driver='GTiff',
+        height=SIZE,
+        width=SIZE,
+        count=channels,
+        crs=crs,
+        dtype=data.dtype,
+        compress='lzw',
+        # transform=transform,
+    ) as dst:
+        dst.write(data)
+
+
+def populate_data(ids: list[str], dir_name: str, with_target: bool = True) -> None:
+    for sid in ids:
+        pre_path = os.path.join(
+            ROOT, DATA_DIR, dir_name, 'pre-event', f'{sid}_pre_disaster.tif'
+        )
+        write_tif(pre_path, channels=3)
+        post_path = os.path.join(
+            ROOT, DATA_DIR, dir_name, 'post-event', f'{sid}_post_disaster.tif'
+        )
+        write_tif(post_path, channels=1)
+        if with_target:
+            target_path = os.path.join(
+                ROOT, DATA_DIR, dir_name, 'target', f'{sid}_building_damage.tif'
+            )
+            write_tif(target_path, channels=1)
+
+
+def main() -> None:
+    make_dirs()
+
+    # Write the ID lists to text files
+    write_list_file(TRAIN_FILE, TRAIN_IDS)
+    write_list_file(HOLDOUT_FILE, HOLDOUT_IDS)
+    write_list_file(VAL_FILE, VAL_IDS)
+
+    # Generate TIF files for the train (with target) and val (no target) splits
+    populate_data(TRAIN_IDS, 'train', with_target=True)
+    populate_data(HOLDOUT_IDS, 'train', with_target=True)
+    populate_data(VAL_IDS, 'val', with_target=False)
+
+    # zip and compute md5
+    zip_filename = os.path.join(ROOT, 'dfc25_track2_trainval')
+    shutil.make_archive(zip_filename, 'zip', ROOT, DATA_DIR)
+
+    def md5(fname: str) -> str:
+        hash_md5 = hashlib.md5()
+        with open(fname, 'rb') as f:
+            for chunk in iter(lambda: f.read(4096), b''):
+                hash_md5.update(chunk)
+        return hash_md5.hexdigest()
+
+    md5sum = md5(zip_filename + '.zip')
+    print(f'MD5 checksum: {md5sum}')
+
+
+if __name__ == '__main__':
+    main()

tests/data/bright/dfc25_track2_trainval.zip

@@ -0,0 +1 @@
+turkey-earthquake_00000413

tests/data/bright/dfc25_track2_trainval/holdout_setlevel.txt

@@ -0,0 +1,3 @@
+bata-explosion_00000049
+bata-explosion_00000014
+bata-explosion_00000047

tests/data/bright/dfc25_track2_trainval/train/post-event/bata-explosion_00000014_post_disaster.tif

@@ -0,0 +1,2 @@
+val-disaster_00000001
+val-disaster_00000002

tests/data/bright/dfc25_track2_trainval/train/post-event/bata-explosion_00000047_post_disaster.tif

@@ -0,0 +1,89 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+# Licensed under the MIT License.
+
+import os
+import shutil
+from pathlib import Path
+
+import matplotlib.pyplot as plt
+import pytest
+import torch
+import torch.nn as nn
+from _pytest.fixtures import SubRequest
+from pytest import MonkeyPatch
+
+from torchgeo.datasets import BRIGHTDFC2025, DatasetNotFoundError
+
+
+class TestBRIGHTDFC2025:
+    @pytest.fixture(params=['train', 'val', 'test'])
+    def dataset(
+        self, monkeypatch: MonkeyPatch, tmp_path: Path, request: SubRequest
+    ) -> BRIGHTDFC2025:
+        md5 = '7b0e24d45fb2d9a4f766196702586414'
+        monkeypatch.setattr(BRIGHTDFC2025, 'md5', md5)
+        url = os.path.join('tests', 'data', 'bright', 'dfc25_track2_trainval.zip')
+        monkeypatch.setattr(BRIGHTDFC2025, 'url', url)
+        root = tmp_path
+        split = request.param
+        transforms = nn.Identity()
+        return BRIGHTDFC2025(root, split, transforms, download=True, checksum=True)
+
+    def test_getitem(self, dataset: BRIGHTDFC2025) -> None:
+        x = dataset[0]
+        assert isinstance(x, dict)
+        assert isinstance(x['image_pre'], torch.Tensor)
+        assert x['image_pre'].shape[0] == 3
+        assert isinstance(x['image_post'], torch.Tensor)
+        assert x['image_post'].shape[0] == 3
+        assert x['image_pre'].shape[-2:] == x['image_post'].shape[-2:]
+        if dataset.split != 'test':
+            assert isinstance(x['mask'], torch.Tensor)
+            assert x['image_pre'].shape[-2:] == x['mask'].shape[-2:]
+
+    def test_len(self, dataset: BRIGHTDFC2025) -> None:
+        if dataset.split == 'train':
+            assert len(dataset) == 3
+        elif dataset.split == 'val':
+            assert len(dataset) == 1
+        else:
+            assert len(dataset) == 2
+
+    def test_already_downloaded(self, dataset: BRIGHTDFC2025) -> None:
+        BRIGHTDFC2025(root=dataset.root)
+
+    def test_not_yet_extracted(self, tmp_path: Path) -> None:
+        filename = 'dfc25_track2_trainval.zip'
+        dir = os.path.join('tests', 'data', 'bright')
+        shutil.copyfile(
+            os.path.join(dir, filename), os.path.join(str(tmp_path), filename)
+        )
+        BRIGHTDFC2025(root=str(tmp_path))
+
+    def test_invalid_split(self) -> None:
+        with pytest.raises(AssertionError):
+            BRIGHTDFC2025(split='foo')
+
+    def test_not_downloaded(self, tmp_path: Path) -> None:
+        with pytest.raises(DatasetNotFoundError, match='Dataset not found'):
+            BRIGHTDFC2025(tmp_path)
+
+    def test_corrupted(self, tmp_path: Path) -> None:
+        with open(os.path.join(tmp_path, 'dfc25_track2_trainval.zip'), 'w') as f:
+            f.write('bad')
+        with pytest.raises(RuntimeError, match='Dataset found, but corrupted.'):
+            BRIGHTDFC2025(root=tmp_path, checksum=True)
+
+    def test_plot(self, dataset: BRIGHTDFC2025) -> None:
+        dataset.plot(dataset[0], suptitle='Test')
+        plt.close()
+
+        if dataset.split != 'test':
+            sample = dataset[0]
+            sample['prediction'] = torch.clone(sample['mask'])
+            dataset.plot(sample, suptitle='Prediction')
+            plt.close()
+
+            del sample['mask']
+            dataset.plot(sample, suptitle='Only Prediction')
+            plt.close()

tests/data/bright/dfc25_track2_trainval/train/post-event/bata-explosion_00000049_post_disaster.tif

@@ -11,6 +11,7 @@
 from .benin_cashews import BeninSmallHolderCashews
 from .bigearthnet import BigEarthNet
 from .biomassters import BioMassters
+from .bright import BRIGHTDFC2025
 from .cabuar import CaBuAr
 from .caffe import CaFFe
 from .cbf import CanadianBuildingFootprints
@@ -152,6 +153,7 @@
 
 __all__ = (
     'ADVANCE',
+    'BRIGHTDFC2025',
     'CDL',
     'COWC',
     'DFC2022',