Source code for torchgeo.datamodules.inria

# Copyright (c) Microsoft Corporation. All rights reserved.
# Licensed under the MIT License.

"""InriaAerialImageLabeling datamodule."""

from typing import Any

import kornia.augmentation as K

from ..datasets import InriaAerialImageLabeling
from ..samplers.utils import _to_tuple
from ..transforms import AugmentationSequential
from ..transforms.transforms import _RandomNCrop
from .geo import NonGeoDataModule


class InriaAerialImageLabelingDataModule(NonGeoDataModule):
    """LightningDataModule implementation for the InriaAerialImageLabeling dataset.

    Uses the train/test splits from the dataset and further splits
    the train split into train/val splits.

    .. versionadded:: 0.3
    """

    def __init__(
        self,
        batch_size: int = 64,
        patch_size: tuple[int, int] | int = 64,
        num_workers: int = 0,
        **kwargs: Any,
    ) -> None:
        """Initialize a new InriaAerialImageLabelingDataModule instance.

        Args:
            batch_size: Size of each mini-batch.
            patch_size: Size of each patch, either ``size`` or ``(height, width)``.
                Should be a multiple of 32 for most segmentation architectures.
            num_workers: Number of workers for parallel data loading.
            **kwargs: Additional keyword arguments passed to
                :class:`~torchgeo.datasets.InriaAerialImageLabeling`.
        """
        super().__init__(InriaAerialImageLabeling, 1, num_workers, **kwargs)

        self.patch_size = _to_tuple(patch_size)

        self.train_aug = AugmentationSequential(
            K.Normalize(mean=self.mean, std=self.std),
            K.RandomHorizontalFlip(p=0.5),
            K.RandomVerticalFlip(p=0.5),
            _RandomNCrop(self.patch_size, batch_size),
            data_keys=['image', 'mask'],
        )
        self.aug = AugmentationSequential(
            K.Normalize(mean=self.mean, std=self.std),
            _RandomNCrop(self.patch_size, batch_size),
            data_keys=['image', 'mask'],
        )
        self.predict_aug = AugmentationSequential(
            K.Normalize(mean=self.mean, std=self.std),
            _RandomNCrop(self.patch_size, batch_size),
            data_keys=['image'],
        )

    def setup(self, stage: str) -> None:
        """Set up datasets.

        Args:
            stage: Either 'fit', 'validate', 'test', or 'predict'.
        """
        if stage in ['fit']:
            self.train_dataset = InriaAerialImageLabeling(split='train', **self.kwargs)
        if stage in ['fit', 'validate']:
            self.val_dataset = InriaAerialImageLabeling(split='val', **self.kwargs)
        if stage in ['predict']:
            # Test set masks are not public, use for prediction instead
            self.predict_dataset = InriaAerialImageLabeling(split='test', **self.kwargs)