Add EMNISTDataModule

.gitignore

@@ -1,3 +1,6 @@
+# IDE Settings files
+.vscode/
+
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]

CHANGELOG.md

@@ -9,6 +9,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- Added `EMNISTDataModule`, `BinaryEMNISTDataModule`, and `BinaryEMNIST` dataset ([#676](https://github.com/PyTorchLightning/lightning-bolts/pull/676))
 
 ### Changed
 

docs/source/datamodules_sklearn.rst

@@ -45,4 +45,3 @@ Automatically generates the train, validation and test splits for a Numpy datase
 They are set up as dataloaders for convenience. Optionally, you can pass in your own validation and test splits.
 
 .. autoclass:: pl_bolts.datamodules.sklearn_datamodule.SklearnDataModule
-   :noindex:

docs/source/datamodules_vision.rst

@@ -5,47 +5,49 @@ The following are pre-built datamodules for computer-vision.
 -------------
 
 Supervised learning
---------------------
+-------------------
 These are standard vision datasets with the train, test, val splits pre-generated in DataLoaders with
 the standard transforms (and Normalization) values
 
+BinaryEMNIST
+^^^^^^^^^^^^
+
+.. autoclass:: pl_bolts.datamodules.binary_emnist_datamodule.BinaryEMNISTDataModule
 
 BinaryMNIST
 ^^^^^^^^^^^
 
 .. autoclass:: pl_bolts.datamodules.binary_mnist_datamodule.BinaryMNISTDataModule
-    :noindex:
 
 CityScapes
 ^^^^^^^^^^
 
 .. autoclass:: pl_bolts.datamodules.cityscapes_datamodule.CityscapesDataModule
-    :noindex:
 
 CIFAR-10
 ^^^^^^^^
 
 .. autoclass:: pl_bolts.datamodules.cifar10_datamodule.CIFAR10DataModule
-    :noindex:
+
+EMNIST
+^^^^^^
+
+.. autoclass:: pl_bolts.datamodules.emnist_datamodule.EMNISTDataModule
 
 FashionMNIST
 ^^^^^^^^^^^^
 
 .. autoclass:: pl_bolts.datamodules.fashion_mnist_datamodule.FashionMNISTDataModule
-    :noindex:
-
 
 Imagenet
 ^^^^^^^^
 
 .. autoclass:: pl_bolts.datamodules.imagenet_datamodule.ImagenetDataModule
-    :noindex:
 
 MNIST
 ^^^^^
 
 .. autoclass:: pl_bolts.datamodules.mnist_datamodule.MNISTDataModule
-    :noindex:
 
 Semi-supervised learning
 ------------------------
@@ -56,10 +58,8 @@ Imagenet (ssl)
 ^^^^^^^^^^^^^^
 
 .. autoclass:: pl_bolts.datamodules.ssl_imagenet_datamodule.SSLImagenetDataModule
-    :noindex:
 
 STL-10
 ^^^^^^
 
 .. autoclass:: pl_bolts.datamodules.stl10_datamodule.STL10DataModule
-    :noindex:

pl_bolts/datamodules/__init__.py

@@ -1,7 +1,9 @@
 from pl_bolts.datamodules.async_dataloader import AsynchronousLoader
+from pl_bolts.datamodules.binary_emnist_datamodule import BinaryEMNISTDataModule
 from pl_bolts.datamodules.binary_mnist_datamodule import BinaryMNISTDataModule
 from pl_bolts.datamodules.cifar10_datamodule import CIFAR10DataModule, TinyCIFAR10DataModule
 from pl_bolts.datamodules.cityscapes_datamodule import CityscapesDataModule
+from pl_bolts.datamodules.emnist_datamodule import EMNISTDataModule
 from pl_bolts.datamodules.experience_source import DiscountedExperienceSource, ExperienceSource, ExperienceSourceDataset
 from pl_bolts.datamodules.fashion_mnist_datamodule import FashionMNISTDataModule
 from pl_bolts.datamodules.imagenet_datamodule import ImagenetDataModule
@@ -33,4 +35,6 @@
     'STL10DataModule',
     'VOCDetectionDataModule',
     'KittiDataset',
+    'EMNISTDataModule',
+    'BinaryEMNISTDataModule',
 ]

pl_bolts/datamodules/binary_emnist_datamodule.py

@@ -0,0 +1,81 @@
+from typing import Any, Optional, Union
+
+from pl_bolts.datamodules.emnist_datamodule import EMNISTDataModule
+from pl_bolts.datasets import BinaryEMNIST
+from pl_bolts.utils import _TORCHVISION_AVAILABLE
+
+
+class BinaryEMNISTDataModule(EMNISTDataModule):
+    """
+    .. figure:: https://user-images.githubusercontent.com/4632336/123210742-4d6b3380-d477-11eb-80da-3e9a74a18a07.png
+        :width: 400
+        :alt: EMNIST
+
+    Please see :class:`~pl_bolts.datamodules.emnist_datamodule.EMNISTDataModule` for more details.
+
+    Example::
+
+        from pl_bolts.datamodules import BinaryEMNISTDataModule
+        dm = BinaryEMNISTDataModule('.')
+        model = LitModel()
+        Trainer().fit(model, datamodule=dm)
+    """
+    name = "binary_emnist"
+    dataset_cls = BinaryEMNIST
+    dims = (1, 28, 28)
+
+    def __init__(
+        self,
+        data_dir: Optional[str] = None,
+        split: str = 'mnist',
+        val_split: Union[int, float] = 0.2,
+        num_workers: int = 16,
+        normalize: bool = False,
+        batch_size: int = 32,
+        seed: int = 42,
+        shuffle: bool = False,
+        pin_memory: bool = False,
+        drop_last: bool = False,
+        strict_val_split: bool = False,
+        *args: Any,
+        **kwargs: Any,
+    ) -> None:
+        """
+        Args:
+            data_dir: Where to save/load the data.
+            split: The dataset has 6 different splits: ``byclass``, ``bymerge``,
+                ``balanced``, ``letters``, ``digits`` and ``mnist``.
+                This argument is passed to :class:`torchvision.datasets.EMNIST`.
+            val_split: Percent (float) or number (int) of samples
+                to use for the validation split.
+            num_workers: How many workers to use for loading data
+            normalize: If ``True``, applies image normalize.
+            batch_size: How many samples per batch to load.
+            seed: Random seed to be used for train/val/test splits.
+            shuffle: If ``True``, shuffles the train data every epoch.
+            pin_memory: If ``True``, the data loader will copy Tensors into
+                CUDA pinned memory before returning them.
+            drop_last: If ``True``, drops the last incomplete batch.
+            strict_val_split: If ``True``, uses the validation split defined in the paper and ignores ``val_split``.
+                Note that it only works with ``"balanced"``, ``"digits"``, ``"letters"``, ``"mnist"`` splits.
+        """
+        if not _TORCHVISION_AVAILABLE:  # pragma: no cover
+            raise ModuleNotFoundError(
+                'You want to use EMNIST dataset loaded from `torchvision` which is not installed yet.'
+            )
+
+        super(BinaryEMNISTDataModule, self).__init__(  # type: ignore[misc]
+            data_dir=data_dir,
+            split=split,
+            val_split=val_split,
+            num_workers=num_workers,
+            normalize=normalize,
+            batch_size=batch_size,
+            seed=seed,
+            shuffle=shuffle,
+            pin_memory=pin_memory,
+            drop_last=drop_last,
+            strict_val_split=strict_val_split,
+            *args,
+            **kwargs,
+        )