glacier ds

docs/api/datamodules.rst

@@ -114,6 +114,11 @@ GID-15
 
 .. autoclass:: GID15DataModule
 
+Glacier Calving Front
+^^^^^^^^^^^^^^^^^^^^^
+
+.. autoclass:: GlacierCalvingFrontDataModule
+
 Inria Aerial Image Labeling
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
 

docs/api/datasets.rst

@@ -307,6 +307,11 @@ GID-15
 
 .. autoclass:: GID15
 
+Glacier Calving Front
+^^^^^^^^^^^^^^^^^^^^^
+
+.. autoclass:: GlacierCalvingFront
+
 IDTReeS
 ^^^^^^^
 

docs/api/datasets/non_geo_datasets.csv

@@ -20,6 +20,7 @@ Dataset,Task,Source,License,# Samples,# Classes,Size (px),Resolution (m),Bands
 `Forest Damage`_,OD,Drone imagery,"CDLA-Permissive-1.0","1,543",4,"1,500x1,500",,RGB
 `GeoNRW`_,S,Aerial,"CC-BY-4.0","7,783",11,"1,000x1,000",1,"RGB, DEM"
 `GID-15`_,S,Gaofen-2,-,150,15,"6,800x7,200",3,RGB
+`GlacierCalvingFront`_,S,"Sentinel-1, TerraSAR-X, TanDEM-X, ENVISAT, European Remote Sensing Satellite 1&2, ALOS PALSAR, and RADARSAT-1","CC-BY-4.0","93,820",4,"256x256",6-20,"SAR"
 `IDTReeS`_,"OD,C",Aerial,"CC-BY-4.0",591,33,200x200,0.1--1,RGB
 `Inria Aerial Image Labeling`_,S,Aerial,-,360,2,"5,000x5,000",0.3,RGB
 `LandCover.ai`_,S,Aerial,"CC-BY-NC-SA-4.0","10,674",5,512x512,0.25--0.5,RGB

tests/conf/glacier_calving_front.yaml

@@ -0,0 +1,16 @@
+model:
+  class_path: SemanticSegmentationTask
+  init_args:
+    loss: 'ce'
+    model: 'unet'
+    backbone: 'resnet18'
+    in_channels: 3
+    num_classes: 11
+    num_filters: 1
+    ignore_index: null
+data:
+  class_path: GlacierCalvingFrontDataModule
+  init_args:
+    batch_size: 1
+  dict_kwargs:
+    root: 'tests/data/glacier_calving_front'

tests/data/glacier_calving_front/data.py

@@ -0,0 +1,75 @@
+import hashlib
+import os
+import shutil
+
+import numpy as np
+from PIL import Image
+
+# Define the root directory and subdirectories
+root_dir = 'glacier_calving_data'
+sub_dirs = ['zones', 'sar_images', 'fronts']
+splits = ['train', 'val', 'test']
+
+zone_file_names = [
+    'Crane_2002-11-09_ERS_20_2_061_zones__93_102_0_0_0.png',
+    'Crane_2007-09-22_ENVISAT_20_1_467_zones__93_102_8_1024_0.png',
+    'JAC_2015-12-23_TSX_6_1_005_zones__57_49_195_384_1024.png',
+]
+
+IMG_SIZE = 32
+
+
+# Function to create dummy images
+def create_dummy_image(path: str, shape: tuple[int], pixel_values: list[int]) -> None:
+    data = np.random.choice(pixel_values, size=shape, replace=True).astype(np.uint8)
+    img = Image.fromarray(data)
+    img.save(path)
+
+
+def create_zone_images(split: str, filename: str) -> None:
+    zone_pixel_values = [0, 64, 127, 255]
+    path = os.path.join(root_dir, 'zones', split, filename)
+    create_dummy_image(path, (IMG_SIZE, IMG_SIZE), zone_pixel_values)
+
+
+def create_sar_images(split: str, filename: str) -> None:
+    sar_pixel_values = range(256)
+    path = os.path.join(root_dir, 'sar_images', split, filename)
+    create_dummy_image(path, (IMG_SIZE, IMG_SIZE), sar_pixel_values)
+
+
+def create_front_images(split: str, filename: str) -> None:
+    sar_pixel_values = range(256)
+    path = os.path.join(root_dir, 'fronts', split, filename)
+    create_dummy_image(path, (IMG_SIZE, IMG_SIZE), sar_pixel_values)
+
+
+if os.path.exists(root_dir):
+    shutil.rmtree(root_dir)
+
+# Create the directory structure
+for sub_dir in sub_dirs:
+    for split in splits:
+        os.makedirs(os.path.join(root_dir, sub_dir, split), exist_ok=True)
+
+# Create dummy data for all splits and filenames
+for split in splits:
+    for filename in zone_file_names:
+        create_zone_images(split, filename)
+        create_sar_images(split, filename.replace('_zones_', '_'))
+        create_front_images(split, filename.replace('_zones_', '_front_'))
+
+# zip and compute md5
+shutil.make_archive(root_dir, 'zip', '.', root_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('glacier_calving_data.zip')
+print(f'MD5 checksum: {md5sum}')

tests/datasets/test_glacier_calving_front.py

@@ -0,0 +1,76 @@
+# 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 DatasetNotFoundError, GlacierCalvingFront
+
+
+class TestGlacierCalvingFront:
+    @pytest.fixture(params=['train', 'test'])
+    def dataset(
+        self, monkeypatch: MonkeyPatch, tmp_path: Path, request: SubRequest
+    ) -> GlacierCalvingFront:
+        md5 = '0b5c05bea31ff666f8eba18b43d4a01f'
+        monkeypatch.setattr(GlacierCalvingFront, 'md5', md5)
+        url = os.path.join(
+            'tests', 'data', 'glacier_calving_front', 'glacier_calving_data.zip'
+        )
+        monkeypatch.setattr(GlacierCalvingFront, 'url', url)
+        root = tmp_path
+        split = request.param
+        transforms = nn.Identity()
+        return GlacierCalvingFront(
+            root, split, transforms, download=True, checksum=True
+        )
+
+    def test_getitem(self, dataset: GlacierCalvingFront) -> None:
+        x = dataset[0]
+        assert isinstance(x, dict)
+        assert isinstance(x['image'], torch.Tensor)
+        assert x['image'].shape[0] == 1
+        assert isinstance(x['mask_zone'], torch.Tensor)
+        assert x['image'].shape[-2:] == x['mask_zone'].shape[-2:]
+
+    def test_len(self, dataset: GlacierCalvingFront) -> None:
+        if dataset.split == 'train':
+            assert len(dataset) == 3
+        else:
+            assert len(dataset) == 3
+
+    def test_already_downloaded(self, dataset: GlacierCalvingFront) -> None:
+        GlacierCalvingFront(root=dataset.root)
+
+    def test_not_yet_extracted(self, tmp_path: Path) -> None:
+        filename = 'glacier_calving_data.zip'
+        dir = os.path.join('tests', 'data', 'glacier_calving_front')
+        shutil.copyfile(
+            os.path.join(dir, filename), os.path.join(str(tmp_path), filename)
+        )
+        GlacierCalvingFront(root=str(tmp_path))
+
+    def test_invalid_split(self) -> None:
+        with pytest.raises(AssertionError):
+            GlacierCalvingFront(split='foo')
+
+    def test_not_downloaded(self, tmp_path: Path) -> None:
+        with pytest.raises(DatasetNotFoundError, match='Dataset not found'):
+            GlacierCalvingFront(tmp_path)
+
+    def test_plot(self, dataset: GlacierCalvingFront) -> None:
+        dataset.plot(dataset[0], suptitle='Test')
+        plt.close()
+
+        sample = dataset[0]
+        sample['prediction'] = torch.clone(sample['mask_zone'])
+        dataset.plot(sample, suptitle='Prediction')
+        plt.close()

tests/trainers/test_segmentation.py

@@ -58,6 +58,7 @@ class TestSemanticSegmentationTask:
             'etci2021',
             'geonrw',
             'gid15',
+            'glacier_calving_front',
             'inria',
             'l7irish',
             'l8biome',

torchgeo/datamodules/__init__.py

@@ -19,6 +19,7 @@
 from .geo import BaseDataModule, GeoDataModule, NonGeoDataModule
 from .geonrw import GeoNRWDataModule
 from .gid15 import GID15DataModule
+from .glacier_calving_front import GlacierCalvingFrontDataModule
 from .inria import InriaAerialImageLabelingDataModule
 from .iobench import IOBenchDataModule
 from .l7irish import L7IrishDataModule
@@ -80,6 +81,7 @@
     'FireRiskDataModule',
     'GeoNRWDataModule',
     'GID15DataModule',
+    'GlacierCalvingFrontDataModule',
     'InriaAerialImageLabelingDataModule',
     'LandCoverAIDataModule',
     'LandCoverAI100DataModule',

torchgeo/datamodules/glacier_calving_front.py

@@ -0,0 +1,67 @@
+# Copyright (c) Microsoft Corporation. All rights reserved.
+# Licensed under the MIT License.
+
+"""GlacierCalvingFront datamodule."""
+
+from typing import Any
+
+import kornia.augmentation as K
+import torch
+
+from ..datasets import GlacierCalvingFront
+from ..transforms import AugmentationSequential
+from .geo import NonGeoDataModule
+
+
+class GlacierCalvingFrontDataModule(NonGeoDataModule):
+    """LightningDataModule implementation for the GlacierCalvingFront dataset.
+
+    Implements the default splits that come with the dataset.
+
+    .. versionadded:: 0.7
+    """
+
+    mean = torch.Tensor([0.5517])
+    std = torch.Tensor([11.8478])
+
+    def __init__(
+        self, batch_size: int = 64, num_workers: int = 0, size: int = 256, **kwargs: Any
+    ) -> None:
+        """Initialize a new GlacierCalvingFrontDataModule instance.
+
+        Args:
+            batch_size: Size of each mini-batch.
+            num_workers: Number of workers for parallel data loading.
+            size: resize images of input size 1000x1000 to size x size
+            **kwargs: Additional keyword arguments passed to
+                :class:`~torchgeo.datasets.GlacierCalvingFront`.
+        """
+        super().__init__(GlacierCalvingFront, batch_size, num_workers, **kwargs)
+
+        self.train_aug = AugmentationSequential(
+            K.Normalize(mean=self.mean, std=self.std),
+            K.Resize(size),
+            K.RandomHorizontalFlip(p=0.5),
+            K.RandomVerticalFlip(p=0.5),
+            data_keys=['image', 'mask'],
+        )
+
+        self.aug = AugmentationSequential(
+            K.Normalize(mean=self.mean, std=self.std),
+            K.Resize(size),
+            data_keys=['image', 'mask'],
+        )
+
+        self.size = size
+
+    def setup(self, stage: str) -> None:
+        """Set up datasets.
+
+        Args:
+            stage: Either 'fit', 'validate', 'test', or 'predict'.
+        """
+        if stage in ['fit', 'validate']:
+            self.train_dataset = GlacierCalvingFront(split='train', **self.kwargs)
+            self.val_dataset = GlacierCalvingFront(split='val', **self.kwargs)
+        if stage in ['test']:
+            self.test_dataset = GlacierCalvingFront(split='test', **self.kwargs)

torchgeo/datasets/__init__.py

@@ -59,6 +59,7 @@
 )
 from .geonrw import GeoNRW
 from .gid15 import GID15
+from .glacier_calving_front import GlacierCalvingFront
 from .globbiomass import GlobBiomass
 from .idtrees import IDTReeS
 from .inaturalist import INaturalist
@@ -223,6 +224,7 @@
     'ForestDamage',
     'GeoNRW',
     'GID15',
+    'GlacierCalvingFront',
     'IDTReeS',
     'InriaAerialImageLabeling',
     'LandCoverAI',