albumentations.core.tensor

CPU Tensor target validation and NumPy bridge helpers.

Members

validate_tensor_input(
    value: torch.Tensor,
    data_name: str,
    canonical_name: str
)

Validate a plain CPU Tensor at the public Compose boundary.

Name	Type	Default	Description
value	torch.Tensor	-	-
data_name	str	-	-
canonical_name	str	-	-

tensor_to_numpy_annotation(
    value: torch.Tensor,
    target: str
)

Return a NumPy view of a validated Tensor annotation matrix.

Name	Type	Default	Description
value	torch.Tensor	-	-
target	str	-	-

tensor_to_numpy_spatial(
    value: torch.Tensor,
    target: str
)

Return a canonical channel-last NumPy view of a canonical Tensor target.

Name	Type	Default	Description
value	torch.Tensor	-	-
target	str	-	-

validate_tensor_metadata_input(
    value: torch.Tensor,
    data_name: str,
    target: str | None
)

Validate a Tensor read through `targets_as_params` before sampling.

Name	Type	Default	Description
value	torch.Tensor	-	-
data_name	str	-	-
target	One of: str None	-	-

tensor_metadata_field_target(
    name: object
)

Return the canonical target for a conventional Tensor metadata field.

Name	Type	Default	Description
name	object	-	-

tensor_metadata_to_numpy(
    value: torch.Tensor,
    target: str | None
)

Convert one validated `targets_as_params` Tensor to the NumPy layout consumed by a transform.

Name	Type	Default	Description
value	torch.Tensor	-	-
target	One of: str None	-	-

numpy_to_tensor_spatial(
    value: NDArray[np.generic],
    target: str
)

Return a canonical Tensor view of a canonical channel-last NumPy target.

Name	Type	Default	Description
value	NDArray[np.generic]	-	-
target	str	-	-

numpy_to_tensor_annotation(
    value: NDArray[np.generic],
    target: str
)

Return a float32 Tensor annotation matrix from a NumPy processor result.

Name	Type	Default	Description
value	NDArray[np.generic]	-	-
target	str	-	-