Skip to content

Classification problem #78

@KirinHao

Description

@KirinHao

I have some problems with the data set used in the classification section.
In the classification experiment, is only the training set and validation set of the VisDA-2017 data set used? Is the test set directly VisDA-2017 validation set? I know little about this dataset and look forward to your reply.
Thank you very much!

    # in cls.vision.datasets.util.py
    elif dataset_name in datasets.__dict__:
        # load datasets from common.vision.datasets
        dataset = datasets.__dict__[dataset_name]

        def concat_dataset(tasks, **kwargs):
            return ConcatDataset([dataset(task=task, **kwargs) for task in tasks])

        train_source_dataset = concat_dataset(root=root, tasks=source, download=True, transform=train_source_transform)
        train_target_dataset = concat_dataset(root=root, tasks=target, download=True, transform=train_target_transform)
        val_dataset = concat_dataset(root=root, tasks=target, download=True, transform=val_transform)
        if dataset_name == 'DomainNet':
            test_dataset = concat_dataset(root=root, tasks=target, split='test', download=True, transform=val_transform)
        else:
            test_dataset = val_dataset  # about here

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions