Source code for common.vision.datasets.resisc45
"""
@author: Junguang Jiang
@contact: [email protected]
"""
from torchvision.datasets.folder import ImageFolder
import random
[docs]class Resisc45(ImageFolder):
"""`Resisc45 <http://www.escience.cn/people/JunweiHan/NWPU-RESISC45.html>`_ dataset \
is a scene classification task from remote sensing images. There are 45 classes, \
containing 700 images each, including tennis court, ship, island, lake, \
parking lot, sparse residential, or stadium. \
The image size is RGB 256x256 pixels.
.. note:: You need to download the source data manually into `root` directory.
Args:
root (str): Root directory of dataset
split (str, optional): The dataset split, supports ``train``, or ``test``.
transform (callable, optional): A function/transform that takes in an PIL image and returns a \
transformed version. E.g, :class:`torchvision.transforms.RandomCrop`.
target_transform (callable, optional): A function/transform that takes in the target and transforms it.
"""
def __init__(self, root, split='train', download=False, **kwargs):
super(Resisc45, self).__init__(root, **kwargs)
random.seed(0)
random.shuffle(self.samples)
if split == 'train':
self.samples = self.samples[:25200]
else:
self.samples = self.samples[25200:]
@property
def num_classes(self) -> int:
"""Number of classes"""
return len(self.classes)
