FedVisionDataset

class fl_sim.data_processing.FedVisionDataset(datadir: str | Path | None = None, transform: str | Callable | None = 'none', seed: int = 0, **extra_config: Any)

基类：FedDataset, ABC

Base class for all federated vision datasets.

Methods that have to be implemented by subclasses:

• get_dataloader

• _preload

• evaluate

Properties that have to be implemented by subclasses:

• url

• candidate_models

• doi

• label_map

参数:

• datadir (Union[pathlib.Path, str], optional) -- Directory to store data. If None, use default directory.

• transform (Union[str, Callable], default "none") -- Transform to apply to data. Conventions: "none" means no transform, using TensorDataset; None for default transform from torchvision.

• seed (int, default 0) -- Random seed for data partitioning.

• **extra_config (dict, optional) -- Extra configurations.

get_class(label: Tensor) → str

Get class name from label.

参数:

label (torch.Tensor) -- Label.

返回:

Class name.

返回类型:

str

get_classes(labels: Tensor) → List[str]

Get class names from labels.

参数:

labels (torch.Tensor) -- Labels.

返回:

Class names.

返回类型:

List[str]

abstract get_dataloader(train_bs: int, test_bs: int, client_idx: int | None = None) → Tuple[DataLoader, DataLoader]

Get dataloader for client client_idx or get global dataloader.

abstract property label_map: dict

Label map for the dataset.

load_partition_data(batch_size: int | None = None) → tuple

Partition data into all local clients.

参数:

batch_size (int, optional) -- Batch size for dataloader. If None, use default batch size.

返回:

• train_clients_num: int

  Number of training clients.

• train_data_num: int

  Number of training data.

• test_data_num: int

  Number of testing data.

• train_data_global: torch.utils.data.DataLoader

  Global training dataloader.

• test_data_global: torch.utils.data.DataLoader

  Global testing dataloader.

• data_local_num_dict: dict

  Number of local training data for each client.

• train_data_local_dict: dict

  Local training dataloader for each client.

• test_data_local_dict: dict

  Local testing dataloader for each client.

• n_class: int

  Number of classes.

返回类型:

tuple

load_partition_data_distributed(process_id: int, batch_size: int | None = None) → tuple

Get local dataloader at client process_id or get global dataloader.

参数:

• process_id (int) -- Index of the client to get dataloader. If None, get the dataloader containing all data, usually used for centralized training.

• batch_size (int, optional) -- Batch size for dataloader. If None, use default batch size.

返回:

• train_clients_num: int

  Number of training clients.

• train_data_num: int

  Number of training data.

• train_data_global: torch.utils.data.DataLoader or None

  Global training dataloader.

• test_data_global: torch.utils.data.DataLoader or None

  Global testing dataloader.

• local_data_num: int

  Number of local training data.

• train_data_local: torch.utils.data.DataLoader or None

  Local training dataloader.

• test_data_local: torch.utils.data.DataLoader or None

  Local testing dataloader.

• n_class: int

  Number of classes.

返回类型:

tuple

property n_class: int

Number of classes.

static show_image(tensor: Tensor | ndarray) → Image

Show image from tensor.

参数:

tensor (Union[torch.Tensor, np.ndarray]) -- Image tensor with shape (C, H, W) or (H, W, C) or (H, W), where C is channel, H is height, W is width. C must be 1 or 3.

返回:

PIL image.

返回类型:

Image.Image