Differences with torchvision.datasets.VOCSegmentation

torchvision.datasets.VOCSegmentation

class torchvision.datasets.VOCSegmentation(
    root: str,
    year: str='2012',
    image_set: str='train',
    download: bool=False,
    transform: Optional[Callable]=None,
    target_transform: Optional[Callable]=None,
    transforms: Optional[Callable]=None
    )

For more information, see torchvision.datasets.VOCSegmentation.

mindspore.dataset.VOCDataset

class mindspore.dataset.VOCDataset(
    dataset_dir,
    task="Segmentation",
    usage="train",
    class_indexing=None,
    num_samples=None,
    num_parallel_workers=None,
    shuffle=None,
    decode=False,
    sampler=None,
    num_shards=None,
    shard_id=None,
    cache=None,
    extra_metadata=False,
    decrypt=None
    )

For more information, see mindspore.dataset.VOCDataset.

Differences

PyTorch: Pascal VOC Segmentation Dataset.

MindSpore: A source dataset for reading and parsing VOC dataset.

Categories	Subcategories	PyTorch	MindSpore	Difference
Parameter	Parameter1	root	dataset_dir	-
	Parameter2	year	-	Not supported by MindSpore
	Parameter3	image_set	usage	-
	Parameter4	download	-	Not supported by MindSpore
	Parameter5	transform	-	Supported by mindspore.dataset.map operation
	Parameter6	target_transform	-	Supported by mindspore.dataset.map operation
	Parameter7	transforms	-	Supported by mindspore.dataset.map operation
	Parameter8	-	task	Set the task type of reading voc data
	Parameter9	-	class_indexing	A str-to-int mapping from label name to index
	Parameter10	-	num_samples	The number of images to be included in the dataset
	Parameter11	-	num_parallel_workers	Number of worker threads to read the data
	Parameter12	-	shuffle	Whether to perform shuffle on the dataset
	Parameter13	-	decode	Whether to decode the images after reading
	Parameter14	-	sampler	Object used to choose samples from the dataset
	Parameter15	-	num_shards	Number of shards that the dataset will be divided into
	Parameter16	-	shard_id	The shard ID within num_shards
	Parameter17	-	cache	Use tensor caching service to speed up dataset processing
	Parameter18	-	extra_metadata	Flag to add extra meta-data to row
	Parameter19	-	decrypt	Image decryption function

Code Example

import mindspore.dataset as ds
import torchvision.transforms as T
import torchvision.datasets as datasets
from torch.utils.data import DataLoader

# In MindSpore, the generated dataset with different task setting has different output columns.

voc_dataset_dir = "/path/to/voc_dataset_directory/"

# task = Segmentation, output columns: [image, dtype=uint8], [target,dtype=uint8].
dataset = ds.VOCDataset(dataset_dir=voc_dataset_dir,
                                                    task="Segmentation",
                                                    usage="train")
for item in dataset:
    print("item:", item[0])
    print(len(item[0]))
    break
# Out:
# item: [255 216 255 ...  73 255 217]
# 52544

# In torch, the output will be result of transform, eg. RandomCrop
root = "/path/to/voc_dataset_directory2/"

dataset = datasets.VOCSegmentation(root, image_set='train', year='2012', transform=T.RandomCrop(300))
print(dataset)
print(type(dataset))