[fbsync] Add SUN397 Dataset (#5132)

Summary:
* dataset class added

* fix code format

* fixed requested changes

* fixed issues in sun397

* Update torchvision/datasets/sun397.py

Reviewed By: sallysyw

Differential Revision: D33479277

fbshipit-source-id: 374d098c261adeacd073fae141380130a6c3aa95

Co-authored-by: Nicolas Hug <[email protected]>
Co-authored-by: Philip Meier <[email protected]>

Author: 3 people

docs/source/datasets.rst

@@ -71,6 +71,7 @@ You can also create your own datasets using the provided :ref:`base classes <bas
     SEMEION
     Sintel
     STL10
+    SUN397
     SVHN
     UCF101
     USPS

test/test_datasets.py

@@ -2206,6 +2206,52 @@ def inject_fake_data(self, tmpdir: str, config):
         return len(sampled_classes * n_samples_per_class)
 
 
+class SUN397TestCase(datasets_utils.ImageDatasetTestCase):
+    DATASET_CLASS = datasets.SUN397
+
+    ADDITIONAL_CONFIGS = datasets_utils.combinations_grid(
+        split=("train", "test"),
+        partition=(1, 10, None),
+    )
+
+    def inject_fake_data(self, tmpdir: str, config):
+        data_dir = pathlib.Path(tmpdir) / "SUN397"
+        data_dir.mkdir()
+
+        num_images_per_class = 5
+        sampled_classes = ("abbey", "airplane_cabin", "airport_terminal")
+        im_paths = []
+
+        for cls in sampled_classes:
+            image_folder = data_dir / cls[0]
+            im_paths.extend(
+                datasets_utils.create_image_folder(
+                    image_folder,
+                    image_folder / cls,
+                    file_name_fn=lambda idx: f"sun_{idx}.jpg",
+                    num_examples=num_images_per_class,
+                )
+            )
+
+        with open(data_dir / "ClassName.txt", "w") as file:
+            file.writelines("\n".join(f"/{cls[0]}/{cls}" for cls in sampled_classes))
+
+        if config["partition"] is not None:
+            num_samples = max(len(im_paths) // (2 if config["split"] == "train" else 3), 1)
+
+            with open(data_dir / f"{config['split'].title()}ing_{config['partition']:02d}.txt", "w") as file:
+                file.writelines(
+                    "\n".join(
+                        f"/{f_path.relative_to(data_dir).as_posix()}"
+                        for f_path in random.choices(im_paths, k=num_samples)
+                    )
+                )
+        else:
+            num_samples = len(im_paths)
+
+        return num_samples
+
+
 class DTDTestCase(datasets_utils.ImageDatasetTestCase):
     DATASET_CLASS = datasets.DTD
     FEATURE_TYPES = (PIL.Image.Image, int)

torchvision/datasets/__init__.py

@@ -28,6 +28,7 @@
 from .sbu import SBU
 from .semeion import SEMEION
 from .stl10 import STL10
+from .sun397 import SUN397
 from .svhn import SVHN
 from .ucf101 import UCF101
 from .usps import USPS
@@ -51,6 +52,7 @@
     "MNIST",
     "KMNIST",
     "STL10",
+    "SUN397",
     "SVHN",
     "PhotoTour",
     "SEMEION",

torchvision/datasets/sun397.py

@@ -0,0 +1,98 @@
+from pathlib import Path
+from typing import Any, Tuple, Callable, Optional
+
+import PIL.Image
+
+from .utils import verify_str_arg, download_and_extract_archive
+from .vision import VisionDataset
+
+
+class SUN397(VisionDataset):
+    """`The SUN397 Data Set <https://vision.princeton.edu/projects/2010/SUN/>`_.
+
+    The SUN397 or Scene UNderstanding (SUN) is a dataset for scene recognition consisting of
+    397 categories with 108'754 images. The dataset also provides 10 partitions for training
+    and testing, with each partition consisting of 50 images per class.
+
+    Args:
+        root (string): Root directory of the dataset.
+        split (string, optional): The dataset split, supports ``"train"`` (default) and ``"test"``.
+        partition (int, optional): A valid partition can be an integer from 1 to 10 or None,
+            for the entire dataset.
+        download (bool, optional): If true, downloads the dataset from the internet and
+            puts it in root directory. If dataset is already downloaded, it is not
+            downloaded again.
+        transform (callable, optional): A function/transform that  takes in an PIL image and returns a transformed
+            version. E.g, ``transforms.RandomCrop``.
+        target_transform (callable, optional): A function/transform that takes in the target and transforms it.
+    """
+
+    _DATASET_URL = "http://vision.princeton.edu/projects/2010/SUN/SUN397.tar.gz"
+    _DATASET_MD5 = "8ca2778205c41d23104230ba66911c7a"
+    _PARTITIONS_URL = "https://vision.princeton.edu/projects/2010/SUN/download/Partitions.zip"
+    _PARTITIONS_MD5 = "29a205c0a0129d21f36cbecfefe81881"
+
+    def __init__(
+        self,
+        root: str,
+        split: str = "train",
+        partition: Optional[int] = 1,
+        download: bool = True,
+        transform: Optional[Callable] = None,
+        target_transform: Optional[Callable] = None,
+    ) -> None:
+        super().__init__(root, transform=transform, target_transform=target_transform)
+        self.split = verify_str_arg(split, "split", ("train", "test"))
+        self.partition = partition
+        self._data_dir = Path(self.root) / "SUN397"
+
+        if self.partition is not None:
+            if self.partition < 0 or self.partition > 10:
+                raise RuntimeError(f"The partition parameter should be an int in [1, 10] or None, got {partition}.")
+
+        if download:
+            self._download()
+
+        if not self._check_exists():
+            raise RuntimeError("Dataset not found. You can use download=True to download it")
+
+        with open(self._data_dir / "ClassName.txt") as f:
+            self.classes = [c[3:].strip() for c in f]
+
+        self.class_to_idx = dict(zip(self.classes, range(len(self.classes))))
+        if self.partition is not None:
+            with open(self._data_dir / f"{self.split.title()}ing_{self.partition:02d}.txt", "r") as f:
+                self._image_files = [self._data_dir.joinpath(*line.strip()[1:].split("/")) for line in f]
+        else:
+            self._image_files = list(self._data_dir.rglob("sun_*.jpg"))
+
+        self._labels = [
+            self.class_to_idx["/".join(path.relative_to(self._data_dir).parts[1:-1])] for path in self._image_files
+        ]
+
+    def __len__(self) -> int:
+        return len(self._image_files)
+
+    def __getitem__(self, idx) -> Tuple[Any, Any]:
+        image_file, label = self._image_files[idx], self._labels[idx]
+        image = PIL.Image.open(image_file).convert("RGB")
+
+        if self.transform:
+            image = self.transform(image)
+
+        if self.target_transform:
+            label = self.target_transform(label)
+
+        return image, label
+
+    def _check_exists(self) -> bool:
+        return self._data_dir.exists() and self._data_dir.is_dir()
+
+    def extra_repr(self) -> str:
+        return "Split: {split}".format(**self.__dict__)
+
+    def _download(self) -> None:
+        if self._check_exists():
+            return
+        download_and_extract_archive(self._DATASET_URL, download_root=self.root, md5=self._DATASET_MD5)
+        download_and_extract_archive(self._PARTITIONS_URL, download_root=str(self._data_dir), md5=self._PARTITIONS_MD5)