dataeval.utils.data.validate_dataset¶

dataeval.utils.data.validate_dataset(dataset, *, expected='any_target', arg_name='dataset', caller=None)¶

Validate that a dataset matches the expected MAITE datum shape.

Parameters:¶

dataset : Any¶

The object passed in as the dataset. Must be Sized and indexable by integer.

expected='any_target'¶

The shape the caller intends to consume. See DatasetKind.

arg_name='dataset'¶

The parameter name to use in error messages.

caller : str, optional¶

Name of the calling function/class — included in error messages for easier debugging. When None, the caller is inferred from the stack.

Returns:¶

The inferred concrete kind. For expected == "any_target" this will be one of "classification" | "object_detection" | "segmentation"; for other inputs it echoes expected.

Return type:¶

DatasetKind

Raises:¶

MaiteShapeError – If the dataset is empty, non-indexable, or its datum shape does not satisfy expected.