OpenImageO#

class torch_uncertainty.datasets.classification.OpenImageO(root, split=None, transform=None, target_transform=None, download=False)[source]#

OpenImage-O dataset.

Parameters:

  • root (str | Path) – Root directory of the datasets.

  • split (str | None) – Unused, for API consistency. Defaults to None.

  • transform (Callable | None) – A function/transform that takes in a PIL image and returns a transformed version. E.g, transforms.RandomCrop. Defaults to None.

  • target_transform (Callable | None) – A function/transform that takes in the target and transforms it. Defaults to None.

  • download (bool) – If True, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again. Defaults to False.

References

[1] Original dataset: The open images dataset v4: Unified image classification, object detection, and visual relationship detection at scale. Kuznetsova, A., et al. The International Journal of Computer Vision.

[2] Curation: ViM: Out-Of-Distribution with Virtual-logit Matching. Wang H., et al. In CVPR 2022.