Skip to content

Data Config

Source

Data configuration.

Float = Annotated[float, PlainSerializer(np_float_to_scientific_str, return_type=str)] module-attribute

Annotated float type, used to serialize floats to strings.

DataConfig

Bases: BaseModel

Data configuration.

If std is specified, mean must be specified as well. Note that setting the std first and then the mean (if they were both None before) will raise a validation error. Prefer instead set_mean_and_std to set both at once. Means and stds are expected to be lists of floats, one for each channel. For supervised tasks, the mean and std of the target could be different from the input data.

All supported transforms are defined in the SupportedTransform enum.

Examples:

Minimum example:

>>> data = DataConfig(
...     data_type="array", # defined in SupportedData
...     patch_size=[128, 128],
...     batch_size=4,
...     axes="YX"
... )

To change the image_means and image_stds of the data:

>>> data.set_means_and_stds(image_means=[214.3], image_stds=[84.5])

One can pass also a list of transformations, by keyword, using the SupportedTransform value:

>>> from careamics.config.support import SupportedTransform
>>> data = DataConfig(
...     data_type="tiff",
...     patch_size=[128, 128],
...     batch_size=4,
...     axes="YX",
...     transforms=[
...         {
...             "name": "XYFlip",
...         }
...     ]
... )

axes instance-attribute

Axes of the data, as defined in SupportedAxes.

batch_size = Field(default=1, ge=1, validate_default=True) class-attribute instance-attribute

Batch size for training.

data_type instance-attribute

Type of input data, numpy.ndarray (array) or paths (tiff, czi, and custom), as defined in SupportedData.

image_means = Field(default=None, min_length=0, max_length=32) class-attribute instance-attribute

Means of the data across channels, used for normalization.

image_stds = Field(default=None, min_length=0, max_length=32) class-attribute instance-attribute

Standard deviations of the data across channels, used for normalization.

patch_size = Field(..., min_length=2, max_length=3) class-attribute instance-attribute

Patch size, as used during training.

target_means = Field(default=None, min_length=0, max_length=32) class-attribute instance-attribute

Means of the target data across channels, used for normalization.

target_stds = Field(default=None, min_length=0, max_length=32) class-attribute instance-attribute

Standard deviations of the target data across channels, used for normalization.

train_dataloader_params = Field(default={'shuffle': True}, validate_default=True) class-attribute instance-attribute

Dictionary of PyTorch training dataloader parameters. The dataloader parameters, should include the shuffle key, which is set to True by default. We strongly recommend to keep it as True to ensure the best training results.

transforms = Field(default=[XYFlipConfig(), XYRandomRotate90Config()], validate_default=True) class-attribute instance-attribute

List of transformations to apply to the data, available transforms are defined in SupportedTransform.

val_dataloader_params = Field(default={}, validate_default=True) class-attribute instance-attribute

Dictionary of PyTorch validation dataloader parameters.

all_elements_power_of_2_minimum_8(patch_list) classmethod

Validate patch size.

Patch size must be powers of 2 and minimum 8.

Parameters:

Name Type Description Default
patch_list list of int

Patch size.

required

Returns:

Type Description
list of int

Validated patch size.

Raises:

Type Description
ValueError

If the patch size is smaller than 8.

ValueError

If the patch size is not a power of 2.

axes_valid(axes) classmethod

Validate axes.

Axes must: - be a combination of 'STCZYX' - not contain duplicates - contain at least 2 contiguous axes: X and Y - contain at most 4 axes - not contain both S and T axes

Parameters:

Name Type Description Default
axes str

Axes to validate.

required

Returns:

Type Description
str

Validated axes.

Raises:

Type Description
ValueError

If axes are not valid.

set_3D(axes, patch_size)

Set 3D parameters.

Parameters:

Name Type Description Default
axes str

Axes.

required
patch_size list of int

Patch size.

required

set_default_pin_memory(dataloader_params) classmethod

Set default pin_memory for dataloader parameters if not provided.

  • If 'pin_memory' is not set, it defaults to True if CUDA is available.

Parameters:

Name Type Description Default
dataloader_params dict of {str: Any}

The dataloader parameters.

required

Returns:

Type Description
dict of {str: Any}

The dataloader parameters with pin_memory default applied.

set_default_train_workers(dataloader_params) classmethod

Set default num_workers for training dataloader if not provided.

  • If 'num_workers' is not set, it defaults to the number of available CPU cores.

Parameters:

Name Type Description Default
dataloader_params dict of {str: Any}

The training dataloader parameters.

required

Returns:

Type Description
dict of {str: Any}

The dataloader parameters with num_workers default applied.

set_means_and_stds(image_means, image_stds, target_means=None, target_stds=None)

Set mean and standard deviation of the data across channels.

This method should be used instead setting the fields directly, as it would otherwise trigger a validation error.

Parameters:

Name Type Description Default
image_means (ndarray, tuple or list)

Mean values for normalization.

required
image_stds (ndarray, tuple or list)

Standard deviation values for normalization.

required
target_means (ndarray, tuple or list)

Target mean values for normalization, by default ().

None
target_stds (ndarray, tuple or list)

Target standard deviation values for normalization, by default ().

None

set_val_workers_to_match_train()

Set validation dataloader num_workers to match training dataloader.

If num_workers is not specified in val_dataloader_params, it will be set to the same value as train_dataloader_params["num_workers"].

Returns:

Type Description
Self

Validated data model with synchronized num_workers.

shuffle_train_dataloader(train_dataloader_params) classmethod

Validate that "shuffle" is included in the training dataloader params.

A warning will be raised if shuffle=False.

Parameters:

Name Type Description Default
train_dataloader_params dict of {str: Any}

The training dataloader parameters.

required

Returns:

Type Description
dict of {str: Any}

The validated training dataloader parameters.

Raises:

Type Description
ValueError

If "shuffle" is not included in the training dataloader params.

std_only_with_mean()

Check that mean and std are either both None, or both specified.

Returns:

Type Description
Self

Validated data model.

Raises:

Type Description
ValueError

If std is not None and mean is None.

validate_dimensions()

Validate 2D/3D dimensions between axes, patch size and transforms.

Returns:

Type Description
Self

Validated data model.

Raises:

Type Description
ValueError

If the transforms are not valid.

np_float_to_scientific_str(x)

Return a string scientific representation of a float.

In particular, this method is used to serialize floats to strings, allowing numpy.float32 to be passed in the Pydantic model and written to a yaml file as str.

Parameters:

Name Type Description Default
x float

Input value.

required

Returns:

Type Description
str

Scientific string representation of the input value.