CIFAR10SuperPixelDataset

class dgl.data.CIFAR10SuperPixelDataset(raw_dir=None, split='train', use_feature=False, force_reload=False, verbose=False, transform=None)

Bases: SuperPixelDataset

CIFAR10 superpixel dataset for the graph classification task.

DGL dataset of CIFAR10 in the benchmark-gnn which contains graphs converted fromt the original CIFAR10 images.

Reference http://arxiv.org/abs/2003.00982

Statistics:

• Train examples: 50,000

• Test examples: 10,000

• Size of dataset images: 32

Parameters:

• raw_dir (str) – Directory to store all the downloaded raw datasets. Default: “~/.dgl/”.

• split (str) – Should be chosen from [“train”, “test”] Default: “train”.

• use_feature (bool) –

  • True: Adj matrix defined from super-pixel locations + features

  • False: Adj matrix defined from super-pixel locations (only)

  Default: False.

• force_reload (bool) – Whether to reload the dataset. Default: False.

• verbose (bool) – Whether to print out progress information. Default: False.

• transform (callable, optional) – A transform that takes in a DGLGraph object and returns a transformed version. The DGLGraph object will be transformed before every access.

Examples

>>> from dgl.data import CIFAR10SuperPixelDataset

>>> # CIFAR10 dataset
>>> train_dataset = CIFAR10SuperPixelDataset(split="train")
>>> len(train_dataset)
50000
>>> graph, label = train_dataset[0]
>>> graph
Graph(num_nodes=123, num_edges=984,
    ndata_schemes={'feat': Scheme(shape=(5,), dtype=torch.float32)}
    edata_schemes={'feat': Scheme(shape=(1,), dtype=torch.float32)}),

>>> # support tensor to be index when transform is None
>>> # see details in __getitem__ function
>>> import torch
>>> idx = torch.tensor([0, 1, 2])
>>> train_dataset_subset = train_dataset[idx]
>>> train_dataset_subset[0]
Graph(num_nodes=123, num_edges=984,
    ndata_schemes={'feat': Scheme(shape=(5,), dtype=torch.float32)}
    edata_schemes={'feat': Scheme(shape=(1,), dtype=torch.float32)}),

__getitem__(idx)

Get the idx-th sample.

Parameters:

idx (int or tensor) – The sample index. 1-D tensor as idx is allowed when transform is None.

Returns:

• (dgl.DGLGraph, Tensor) – Graph with node feature stored in feat field and its label.

• or

• dgl.data.utils.Subset – Subset of the dataset at specified indices

__len__()

The number of examples in the dataset.