You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
105 lines
4.4 KiB
105 lines
4.4 KiB
5 months ago
|
from pathlib import Path
|
||
|
|
from typing import Any, Callable, Optional, Tuple, Union
|
||
|
|
|
||
|
|
from PIL import Image
|
||
|
|
|
||
|
|
from .folder import find_classes, make_dataset
|
||
|
|
from .utils import download_and_extract_archive, verify_str_arg
|
||
|
|
from .vision import VisionDataset
|
||
|
|
|
||
|
|
|
||
|
|
class Imagenette(VisionDataset):
|
||
|
|
"""`Imagenette <https://github.com/fastai/imagenette#imagenette-1>`_ image classification dataset.
|
||
|
|
|
||
|
|
Args:
|
||
|
|
root (str or ``pathlib.Path``): Root directory of the Imagenette dataset.
|
||
|
|
split (string, optional): The dataset split. Supports ``"train"`` (default), and ``"val"``.
|
||
|
|
size (string, optional): The image size. Supports ``"full"`` (default), ``"320px"``, and ``"160px"``.
|
||
|
|
download (bool, optional): If ``True``, downloads the dataset components and places them in ``root``. Already
|
||
|
|
downloaded archives are not downloaded again.
|
||
|
|
transform (callable, optional): A function/transform that takes in a 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.
|
||
|
|
|
||
|
|
Attributes:
|
||
|
|
classes (list): List of the class name tuples.
|
||
|
|
class_to_idx (dict): Dict with items (class name, class index).
|
||
|
|
wnids (list): List of the WordNet IDs.
|
||
|
|
wnid_to_idx (dict): Dict with items (WordNet ID, class index).
|
||
|
|
"""
|
||
|
|
|
||
|
|
_ARCHIVES = {
|
||
|
|
"full": ("https://s3.amazonaws.com/fast-ai-imageclas/imagenette2.tgz", "fe2fc210e6bb7c5664d602c3cd71e612"),
|
||
|
|
"320px": ("https://s3.amazonaws.com/fast-ai-imageclas/imagenette2-320.tgz", "3df6f0d01a2c9592104656642f5e78a3"),
|
||
|
|
"160px": ("https://s3.amazonaws.com/fast-ai-imageclas/imagenette2-160.tgz", "e793b78cc4c9e9a4ccc0c1155377a412"),
|
||
|
|
}
|
||
|
|
_WNID_TO_CLASS = {
|
||
|
|
"n01440764": ("tench", "Tinca tinca"),
|
||
|
|
"n02102040": ("English springer", "English springer spaniel"),
|
||
|
|
"n02979186": ("cassette player",),
|
||
|
|
"n03000684": ("chain saw", "chainsaw"),
|
||
|
|
"n03028079": ("church", "church building"),
|
||
|
|
"n03394916": ("French horn", "horn"),
|
||
|
|
"n03417042": ("garbage truck", "dustcart"),
|
||
|
|
"n03425413": ("gas pump", "gasoline pump", "petrol pump", "island dispenser"),
|
||
|
|
"n03445777": ("golf ball",),
|
||
|
|
"n03888257": ("parachute", "chute"),
|
||
|
|
}
|
||
|
|
|
||
|
|
def __init__(
|
||
|
|
self,
|
||
|
|
root: Union[str, Path],
|
||
|
|
split: str = "train",
|
||
|
|
size: str = "full",
|
||
|
|
download=False,
|
||
|
|
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", "val"])
|
||
|
|
self._size = verify_str_arg(size, "size", ["full", "320px", "160px"])
|
||
|
|
|
||
|
|
self._url, self._md5 = self._ARCHIVES[self._size]
|
||
|
|
self._size_root = Path(self.root) / Path(self._url).stem
|
||
|
|
self._image_root = str(self._size_root / self._split)
|
||
|
|
|
||
|
|
if download:
|
||
|
|
self._download()
|
||
|
|
elif not self._check_exists():
|
||
|
|
raise RuntimeError("Dataset not found. You can use download=True to download it.")
|
||
|
|
|
||
|
|
self.wnids, self.wnid_to_idx = find_classes(self._image_root)
|
||
|
|
self.classes = [self._WNID_TO_CLASS[wnid] for wnid in self.wnids]
|
||
|
|
self.class_to_idx = {
|
||
|
|
class_name: idx for wnid, idx in self.wnid_to_idx.items() for class_name in self._WNID_TO_CLASS[wnid]
|
||
|
|
}
|
||
|
|
self._samples = make_dataset(self._image_root, self.wnid_to_idx, extensions=".jpeg")
|
||
|
|
|
||
|
|
def _check_exists(self) -> bool:
|
||
|
|
return self._size_root.exists()
|
||
|
|
|
||
|
|
def _download(self):
|
||
|
|
if self._check_exists():
|
||
|
|
raise RuntimeError(
|
||
|
|
f"The directory {self._size_root} already exists. "
|
||
|
|
f"If you want to re-download or re-extract the images, delete the directory."
|
||
|
|
)
|
||
|
|
|
||
|
|
download_and_extract_archive(self._url, self.root, md5=self._md5)
|
||
|
|
|
||
|
|
def __getitem__(self, idx: int) -> Tuple[Any, Any]:
|
||
|
|
path, label = self._samples[idx]
|
||
|
|
image = Image.open(path).convert("RGB")
|
||
|
|
|
||
|
|
if self.transform is not None:
|
||
|
|
image = self.transform(image)
|
||
|
|
|
||
|
|
if self.target_transform is not None:
|
||
|
|
label = self.target_transform(label)
|
||
|
|
|
||
|
|
return image, label
|
||
|
|
|
||
|
|
def __len__(self) -> int:
|
||
|
|
return len(self._samples)
|