dacapo.experiments.datasplits.datasets

Subpackages

• dacapo.experiments.datasplits.datasets.arrays
• dacapo.experiments.datasplits.datasets.graphstores

Submodules

• dacapo.experiments.datasplits.datasets.dataset
• dacapo.experiments.datasplits.datasets.dataset_config
• dacapo.experiments.datasplits.datasets.dummy_dataset
• dacapo.experiments.datasplits.datasets.dummy_dataset_config
• dacapo.experiments.datasplits.datasets.raw_gt_dataset
• dacapo.experiments.datasplits.datasets.raw_gt_dataset_config

Classes

Dataset	A class to represent a dataset.
DatasetConfig	A class used to define configuration for datasets. This provides the
DummyDataset	DummyDataset is a child class of the Dataset. This class has property 'raw' of Array type and a name.
DummyDatasetConfig	A dummy configuration class for test datasets.
RawGTDataset	A dataset that contains raw and ground truth data. Optionally, it can also contain a mask.
RawGTDatasetConfig	This is a configuration class for the standard dataset with both raw and GT Array.

Package Contents

class dacapo.experiments.datasplits.datasets.Dataset

A class to represent a dataset.

name

The name of the dataset.

Type:

str

raw

The raw dataset.

Type:

Array

gt

The ground truth data.

Type:

Array, optional

mask

The mask for the data.

Type:

Array, optional

weight

The weight of the dataset.

Type:

int, optional

sample_points

The list of sample points in the dataset.

Type:

list[Coordinate], optional

__eq__(other)

Overloaded equality operator for dataset objects.

__hash__()

Calculates a hash for the dataset.

__repr__()

Returns the official string representation of the dataset object.

__str__()

Returns the string representation of the dataset object.

_neuroglancer_layers(prefix='', exclude_layers=None)

Generates neuroglancer layers for raw, gt and mask if they can be viewed by neuroglance, excluding those in the exclude_layers.

Notes

This class is a base class and should not be instantiated.

name: str

raw: dacapo.experiments.datasplits.datasets.arrays.Array

gt: dacapo.experiments.datasplits.datasets.arrays.Array | None

mask: dacapo.experiments.datasplits.datasets.arrays.Array | None

weight: int | None

sample_points: List[funlib.geometry.Coordinate] | None

class dacapo.experiments.datasplits.datasets.DatasetConfig

A class used to define configuration for datasets. This provides the framework to create a Dataset instance.

name

str (eg: “sample_dataset”). A unique identifier to name the dataset. It aids in easy identification and reusability of this dataset. Advised to keep it short and refrain from using special characters.

weight

int (default=1). A numeric value that indicates how frequently this dataset should be sampled in comparison to others. Higher the weight, more frequently it gets sampled.

verify()

Checks and validates the dataset configuration. The specific rules for validation need to be defined by the user.

Notes

This class is used to create a configuration object for datasets.

name: str

weight: int

verify() → Tuple[bool, str]

Method to verify the dataset configuration.

Since there is no specific validation logic defined for this DataSet, this method will always return True as default reaction and a message stating the lack of validation.

Returns:

A tuple of boolean value indicating the check (True or False) and message specifying result of validation.

Return type:

tuple

Raises:

NotImplementedError – If the method is not implemented in the derived class.

Examples

>>> dataset_config = DatasetConfig(name="sample_dataset")
>>> dataset_config.verify()
(True, "No validation for this DataSet")

Notes

This method is used to validate the configuration of the dataset.

class dacapo.experiments.datasplits.datasets.DummyDataset(dataset_config)

DummyDataset is a child class of the Dataset. This class has property ‘raw’ of Array type and a name.

raw

Array The raw data.

__init__(dataset_config)

Initializes the array type ‘raw’ and name for the DummyDataset instance.

Notes

This class is used to create a dataset with raw data.

raw: dacapo.experiments.datasplits.datasets.arrays.Array

name

class dacapo.experiments.datasplits.datasets.DummyDatasetConfig

A dummy configuration class for test datasets.

dataset_type

Clearly mentions the type of dataset

raw_config

This attribute holds the configurations related to dataset arrays.

verify()

A dummy verification method for testing purposes, always returns False and a message.

Notes

This class is used to create a configuration object for the dummy dataset.

dataset_type

raw_config: dacapo.experiments.datasplits.datasets.arrays.ArrayConfig

verify() → Tuple[bool, str]

A dummy method that always indicates the dataset config is not valid.

Returns:

A tuple of False and a message indicating the invalidity.

Raises:

NotImplementedError – If the method is not implemented in the derived class.

Examples

>>> dataset_config = DummyDatasetConfig(raw_config=DummyArrayConfig(name="dummy_array"))
>>> dataset_config.verify()
(False, "This is a DummyDatasetConfig and is never valid")

Notes

This method is used to validate the configuration of the dataset.

class dacapo.experiments.datasplits.datasets.RawGTDataset(dataset_config)

A dataset that contains raw and ground truth data. Optionally, it can also contain a mask.

raw

Array The raw data.

gt

Array The ground truth data.

mask

Optional[Array] The mask data.

sample_points

Optional[List[Coordinate]] The sample points in the graph.

weight

Optional[float] The weight of the dataset.

__init__(dataset_config)

Initialize the dataset.

Notes

This class is a base class and should not be instantiated.

raw: dacapo.experiments.datasplits.datasets.arrays.Array

gt: dacapo.experiments.datasplits.datasets.arrays.Array

mask: dacapo.experiments.datasplits.datasets.arrays.Array | None

sample_points: List[funlib.geometry.Coordinate] | None

name

weight

class dacapo.experiments.datasplits.datasets.RawGTDatasetConfig

This is a configuration class for the standard dataset with both raw and GT Array.

The configuration includes array configurations for raw data, ground truth data and mask data. The configuration for ground truth (GT) data is mandatory, whereas configurations for raw and mask data are optional. It also includes an optional list of points around which training samples will be extracted.

dataset_type

The type of dataset that is being configured.

Type:

class

raw_config

Configuration for the raw data associated with this dataset.

Type:

Optional[ArrayConfig]

gt_config

Configuration for the ground truth data associated with this dataset.

Type:

Optional[ArrayConfig]

mask_config

An optional mask configuration that sets the loss equal to zero on voxels where the mask is 1.

Type:

Optional[ArrayConfig]

sample_points

An optional list of points around which training samples will be extracted.

Type:

Optional[List[Coordinate]]

verify()

A method to verify the validity of the configuration.

Notes

This class is used to create a configuration object for the standard dataset with both raw and GT Array.

dataset_type

raw_config: dacapo.experiments.datasplits.datasets.arrays.ArrayConfig | None

gt_config: dacapo.experiments.datasplits.datasets.arrays.ArrayConfig | None

mask_config: dacapo.experiments.datasplits.datasets.arrays.ArrayConfig | None

sample_points: List[funlib.geometry.Coordinate] | None