Data layers

See Data layers for an introduction to data layers.

Layer

Layer(
    name: str = "",
    data: Any = None,
    meta: Optional[Dict] = None,
    position: Optional[Tuple] = None,
    tile_meta: Optional[TileMeta] = None,
    description: str = "",
    merger: str = "default",
    **meta_kwargs,
)

Base class of data layers.

A data layer holds a single piece of data (an image, a mask, a numeric value, etc.) together with metadata. Layers represent both the inputs and the outputs of algorithms.

Parameters:

Name	Type	Description	Default
name	str	Name of the layer.	""
data	object	Data held by the layer. Its type depends on the subclass.	None
meta	dict	Metadata dictionary. It is merged with description, merger, position, and extra keyword arguments (explicit meta keys take precedence).	None
position	tuple	Position of the layer in global pixel coordinates.	None
tile_meta	TileMeta	Metadata about the layer's position and role in a set of tiles.	None
description	str	Description of the layer, displayed on the algorithm documentation page.	""
merger	str	Strategy used to merge the layer when it is assembled from tiles. Only Mask supports another strategy: "instances".	"default"
**meta_kwargs		Extra metadata, added to meta. In Napari, metadata keys are applied as properties of the displayed layer (e.g. colormap="viridis").	{}

Attributes:

Name	Type	Description
data	Any	Data held by the layer.
name	str	Name of the layer.
meta	dict	Metadata about the layer.
kind	str	A short string identifying the layer type, e.g. "mask".
position	tuple	Position of the layer in global pixel coordinates.
extent	Domain	Region covered by the data in global pixel coordinates. For objects such as points, it is the bounding box of the objects.
size	tuple	Size of the extent.
coords_min	tuple	Lower corner of the extent.
coords_max	tuple	Upper corner of the extent.
ndim	int	Number of spatial dimensions.
shape	tuple	Shape of the data, if it is array-like.
tile_meta	TileMeta	Metadata about the layer's position and role in a set of tiles.

select

select(domain: Domain) -> Layer

Select the part of the layer inside a domain.

Parameters:

Name	Type	Description	Default
domain	Domain	The region to select, in global pixel coordinates.	required

Returns:

Type	Description
Layer	A new layer with the selected data, positioned in global coordinates.

Image

Image(
    data: Optional[ndarray] = None,
    name="Image",
    description="Input image",
    dimensionality: Optional[List[int]] = None,
    rgb: bool = False,
    channel_axis: Optional[int] = None,
    **kwargs,
)

Bases: Layer

Data layer for images: 2D or 3D arrays, optionally multichannel or RGB.

Parameters:

Name	Type	Description	Default
data	ndarray	The image array.	None
name	str	Name of the layer.	"Image"
description	str	Description of the layer, displayed on the algorithm documentation page.	"Input image"
dimensionality	list of int	Accepted numbers of dimensions, for example [2, 3]. By default, any number of dimensions is accepted.	None
rgb	bool	Set to True for RGB images.	False
channel_axis	int	Index of the channel axis, for multichannel images. The channel axis is not counted as a spatial dimension (it doesn't affect ndim and extent), and it is never split into tiles. With rgb=True, the last axis is always the channel axis.	None
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Examples:

>>> image = sk.Image(np.zeros((100, 200)), name="Blurred", colormap="viridis")

Mask

Mask(
    data: Optional[ndarray] = None,
    name: str = "Mask",
    description: str = "Segmentation mask",
    dimensionality: Optional[List[int]] = None,
    channel_axis: Optional[int] = None,
    **kwargs,
)

Bases: Layer

Data layer for segmentation masks.

Masks are label images, where integer values encode either object classes (e.g. for pixel classification) or object instances.

Parameters:

Name	Type	Description	Default
data	ndarray	The label image. Boolean arrays are accepted for binary masks.	None
name	str	Name of the layer.	"Mask"
description	str	Description of the layer, displayed on the algorithm documentation page.	"Segmentation mask"
dimensionality	list of int	Accepted numbers of dimensions, for example [2, 3]. By default, any number of dimensions is accepted.	None
channel_axis	int	Index of the channel axis, if any. The channel axis is not counted as a spatial dimension, and it is never split into tiles.	None
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties. Use merger="instances" for instance segmentation masks, so that labels stay unique when the mask is assembled from tiles.	{}

Examples:

>>> mask = sk.Mask(labels, name="Nuclei", merger="instances")

Points

Points(
    data: Optional[ndarray] = None,
    name="Points",
    description="Input points (2D, 3D)",
    dimensionality: Optional[List[int]] = None,
    **kwargs,
)

Bases: Layer

Data layer for sets of points (2D, 3D).

Parameters:

Name	Type	Description	Default
data	ndarray	Point coordinates, as an array of shape (N, D), where D is the number of dimensions.	None
name	str	Name of the layer.	"Points"
description	str	Description of the layer, displayed on the algorithm documentation page.	"Input points (2D, 3D)"
dimensionality	list of int	Accepted numbers of dimensions, for example [2, 3]. By default, any number of dimensions is accepted.	None
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Vectors

Vectors(
    data: Optional[ndarray] = None,
    name="Vectors",
    description="Input vectors (2D, 3D)",
    dimensionality: Optional[List[int]] = None,
    **kwargs,
)

Bases: Layer

Data layer for sets of vectors (2D, 3D).

Each vector is defined by an origin point and a displacement.

Parameters:

Name	Type	Description	Default
data	ndarray	An array of shape (N, 2, D), where D is the number of dimensions. data[:, 0, :] holds the origins of the vectors, and data[:, 1, :] the displacements from the origins.	None
name	str	Name of the layer.	"Vectors"
description	str	Description of the layer, displayed on the algorithm documentation page.	"Input vectors (2D, 3D)"
dimensionality	list of int	Accepted numbers of dimensions, for example [2, 3]. By default, any number of dimensions is accepted.	None
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Boxes

Boxes(
    data: Optional[ndarray] = None,
    name="Boxes",
    description="Bounding boxes",
    dimensionality: Optional[List[int]] = None,
    **kwargs,
)

Bases: Layer

Data layer for bounding boxes (2D, 3D).

Boxes can be oriented (OBB).

Parameters:

Name	Type	Description	Default
data	ndarray	The coordinates of the four corners of each box, as an array of shape (N, 4, D), where D is the number of dimensions.	None
name	str	Name of the layer.	"Boxes"
description	str	Description of the layer, displayed on the algorithm documentation page.	"Bounding boxes"
dimensionality	list of int	Accepted numbers of dimensions, for example [2, 3]. By default, any number of dimensions is accepted.	None
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Paths

Paths(
    data: Optional[List] = None,
    name="Paths",
    description="Input paths (2D, 3D)",
    dimensionality: Optional[List[int]] = None,
    serializer: str = "default",
    **kwargs,
)

Bases: Layer

Data layer for paths (2D, 3D), for example spline curves.

Parameters:

Name	Type	Description	Default
data	list of numpy.ndarray	One array per path, each of shape (N, D), where N is the number of points in the path and D the number of dimensions.	None
name	str	Name of the layer.	"Paths"
description	str	Description of the layer, displayed on the algorithm documentation page.	"Input paths (2D, 3D)"
dimensionality	list of int	Accepted numbers of dimensions, for example [2, 3]. By default, any number of dimensions is accepted.	None
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Tracks

Tracks(
    data: Optional[ndarray] = None,
    name="Tracks",
    description="Input tracks (2D, 3D)",
    dimensionality: Optional[List[int]] = None,
    **kwargs,
)

Bases: Layer

Data layer for object tracks: point detections linked over time by a track ID.

Parameters:

Name	Type	Description	Default
data	ndarray	An array of shape (N, D+1), with columns [ID, T, (Z), Y, X].	None
name	str	Name of the layer.	"Tracks"
description	str	Description of the layer, displayed on the algorithm documentation page.	"Input tracks (2D, 3D)"
dimensionality	list of int	Accepted numbers of dimensions, for example [2, 3]. By default, any number of dimensions is accepted.	None
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Float

Float(
    data: Optional[float] = None,
    name="Float",
    min: float = float(min),
    max: float = float(max),
    step: float = 0.1,
    default: float = 0.0,
    required: bool = True,
    auto_call: bool = False,
    **kwargs,
)

Bases: Layer

Data layer for floating-point values.

Parameters:

Name	Type	Description	Default
data	float	The value.	None
name	str	Name of the layer.	"Float"
min	float	Minimum accepted value.	float(min)
max	float	Maximum accepted value.	float(max)
step	float	Step size of the spin box in user interfaces.	0.1
default	float	Default value, used when data is not provided.	0.0
auto_call	bool	Re-run the algorithm when the value changes in user interfaces.	False
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Examples:

>>> sigma = sk.Float(name="Sigma", min=0, default=1.0, step=0.5)

Integer

Integer(
    data: Optional[int] = None,
    name="Int",
    default: int = 0,
    required: bool = True,
    auto_call: bool = False,
    min: int = int(min),
    max: int = int(max),
    step: int = 1,
    **kwargs,
)

Bases: Layer

Data layer for integer values.

Parameters:

Name	Type	Description	Default
data	int	The value.	None
name	str	Name of the layer.	"Int"
min	int	Minimum accepted value.	int(min)
max	int	Maximum accepted value.	int(max)
step	int	Step size of the spin box in user interfaces.	1
default	int	Default value, used when data is not provided.	0
auto_call	bool	Re-run the algorithm when the value changes in user interfaces.	False
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Examples:

>>> threshold = sk.Integer(name="Threshold", min=0, max=255, default=128)

Bool

Bool(
    data: Optional[bool] = None,
    name="Bool",
    default: bool = False,
    required: bool = True,
    auto_call: bool = False,
    **kwargs,
)

Bases: Layer

Data layer for boolean values, shown as a checkbox in user interfaces.

Parameters:

Name	Type	Description	Default
data	bool	The value.	None
name	str	Name of the layer.	"Bool"
default	bool	Default value, used when data is not provided.	False
auto_call	bool	Re-run the algorithm when the value changes in user interfaces.	False
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

String

String(
    data: Optional[str] = None,
    name="String",
    default: str = "",
    required: bool = True,
    **kwargs,
)

Bases: Layer

Data layer for strings of text.

Parameters:

Name	Type	Description	Default
data	str	The text.	None
name	str	Name of the layer.	"String"
default	str	Default value, used when data is not provided.	""
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Choice

Choice(
    data: Optional[str] = None,
    name="Choice",
    items: Optional[List] = None,
    required: bool = True,
    default: str = "",
    auto_call: bool = False,
    **kwargs,
)

Bases: Layer

Data layer for a choice among items, shown as a dropdown in user interfaces.

Can be used to represent labels for classification.

Parameters:

Name	Type	Description	Default
data	str	The selected item.	None
name	str	Name of the layer.	"Choice"
items	list of str	The available items.	None
default	str	Default item, used when data is not provided.	""
auto_call	bool	Re-run the algorithm when the selection changes in user interfaces.	False
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Examples:

>>> mode = sk.Choice(items=["reflect", "constant"], default="reflect")

Notification

Notification(
    data: Optional[str] = None,
    level: Optional[str] = "info",
    name="Notification",
    required: bool = True,
    default: str = "",
    **kwargs,
)

Bases: Layer

Data layer for text notifications.

Notifications are printed to the terminal, or shown in user interfaces (as Napari notifications, for example).

Parameters:

Name	Type	Description	Default
data	str	The notification text.	None
level	str	Notification level: "info", "warning", or "error".	"info"
name	str	Name of the layer.	"Notification"
**kwargs		Passed to Layer.	{}

Examples:

>>> notif = sk.Notification("Warning!", level="warning")

Progress

Progress(
    data: Optional[int] = None,
    max_val: Optional[int] = 1,
    name="Progress",
    **kwargs,
)

Bases: Layer

Data layer for progress bars.

Progress is shown in the terminal, or as a progress bar in user interfaces.

Parameters:

Name	Type	Description	Default
data	int	Number of completed steps.	None
max_val	int	Total number of steps.	1
name	str	Name of the layer.	"Progress"
**kwargs		Passed to Layer.	{}

Examples:

>>> max_val = 10
>>> for k in range(max_val):
...     yield sk.Progress(k, max_val=max_val)

Null

Null(
    data: Optional[Any] = None,
    name="None",
    default=None,
    serializer: str = "default",
    **kwargs,
)

Bases: Layer

Data layer for None, or the absence of data.

Parameters:

Name	Type	Description	Default
data	None	Always None; accepted for consistency with other layers.	None
name	str	Name of the layer.	"None"
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}

Any

Any(
    data: Optional[Any] = None,
    name="Any",
    default=None,
    serializer: str = "default",
    **kwargs,
)

Bases: Layer

Data layer for any other kind of data, such as custom Python objects.

Parameters with types that cannot be resolved, and outputs of unrecognized types, are represented by this layer. They are not shown in user interfaces.

Since this layer can hold arbitrary objects, it is generally not serializable, and only works locally (not with servers). Exceptions are simple JSON-compatible types (int, float, bool, str, numpy.ndarray), and lists, tuples, or dictionaries of these types.

Parameters:

Name	Type	Description	Default
data	object	Any kind of data.	None
name	str	Name of the layer.	"Any"
**kwargs		Passed to Layer, e.g. position, meta, or extra metadata such as display properties (colormap="viridis").	{}