renumics.spotlight.dtypes¶
Spotlight data types.
The most dtypes are non-customazable and can be used through simple importing
the respective module variables (e.g. float_dtype,
image_dtype).
Some dtypes are customazable and only their default versions can be defined
through the respective module variables (e.g. category_dtype,
embedding_dtype). For more info, see the module classes.
In the most usage cases string or object aliases can be used instead of the default dtypes. For more info, see the module classes.
The main usage of the dtypes is customizing the spotlight.show.
module-attribute
¶
datetime_dtype = DType('datetime')
Datetime dtype. Aliases: "datetime", datetime.datetime.
module-attribute
¶
category_dtype = CategoryDType()
Categorical dtype with arbitraty categories. Aliases: "Category".
module-attribute
¶
window_dtype = DType('Window')
A single window. Aliases: "Window".
A single window is represented by an array-like with two timestamps in seconds as float values.
Single NaN, infinity and out-of-bound values will be clipped by the audio/time series bounds. In case of both non-valid values no window will be showed. Descending values will be visually highlighted.
module-attribute
¶
bounding_box_dtype = DType('BoundingBox')
Single or multiple bounding boxes. Aliases: "BoundingBox".
A single bounding box is represented by an array-like with its relative coordinates [x_min, y_min, x_max, y_max] (float values scaled onto 0 to 1). Top-left image corner is assumed to be (0, 0).
module-attribute
¶
bounding_boxes_dtype = SequenceDType(bounding_box_dtype)
Multiple bounding boxes.
Multiple bounding boxes for the same image are represented by an array-like of floats with shape (n, 4) (preferred) or (4, n).
module-attribute
¶
embedding_dtype = EmbeddingDType()
Embedding dtype. Aliases: "Embedding", renumics.spotlight.media.Embedding.
module-attribute
¶
array_dtype = ArrayDType()
numpy array dtype. Aliases: "array", np.ndarray.
module-attribute
¶
image_dtype = DType('Image')
Image dtype. Aliases: "Image", renumics.spotlight.media.Image.
module-attribute
¶
audio_dtype = DType('Audio')
Audio dtype. Aliases: "Audio", renumics.spotlight.media.Audio.
module-attribute
¶
mesh_dtype = DType('Mesh')
Mesh dtype. Aliases: "Mesh", renumics.spotlight.media.Mesh.
module-attribute
¶
sequence_1d_dtype = Sequence1DDType()
1D-sequence dtype with arbitraty axis labels. Aliases: "Sequence1D",
renumics.spotlight.media.Sequence1D.
module-attribute
¶
video_dtype = DType('Video')
Video dtype. Aliases: "video", renumics.spotlight.media.Video.
module-attribute
¶
unknown_dtype = DType('unknown')
Unknown dtype. Aliases: "unknown".
module-attribute
¶
any_dtype = DType('any')
Unknown dtype which accepts any value. Aliases: "any".
module-attribute
¶
file_dtype = DType('file')
File Dtype (bytes or str(path)). Aliases: "file".
Bases: DType
Categorical dtype with predefined categories.
Category names and codes are assured to be unique. Empty categories mean to
be defined later (in this case, equivalent to the
category_dtype module variable).
Using with category names
from renumics.spotlight import dtypes dtype = dtypes.CategoryDType(["cat", "dog"]) str(dtype) 'Category' dtype.categories {'cat': 0, 'dog': 1} dtype.inverted_categories
Example of usage with category mapping
from renumics.spotlight import dtypes dtype = dtypes.CategoryDType({"four": 4, "two": 2}) dtype.categories
Example of usage with empty categories
from renumics.spotlight import dtypes dtype = dtypes.CategoryDType() dtype == dtypes.category_dtype True
Bases: DType
Array dtype with optional shape.
Attributes:
| Name | Type | Description |
|---|---|---|
shape |
Optional[Tuple[Optional[int], ...]]
|
Array dimensions. Can be fully or partially defined. No shape means to be defined later. |
Example of usage with defined shape
from renumics.spotlight import dtypes dtype = dtypes.ArrayDType((10, None, 4)) str(dtype) 'array' dtype.shape (10, None, 4) dtype.ndim 3
Example of usage with empty shape
from renumics.spotlight import dtypes dtype = dtypes.ArrayDType() dtype.ndim 0 dtype == dtypes.array_dtype True
Bases: DType
Embedding dtype with optional length.
Attributes:
| Name | Type | Description |
|---|---|---|
length |
Optional[int]
|
Embedding length. No length means to be defined later. |
Example of usage with defined length
from renumics.spotlight import dtypes dtype = dtypes.EmbeddingDType(8) str(dtype) 'Embedding' dtype.length 8
Example of usage with empty length
from renumics.spotlight import dtypes dtype = dtypes.EmbeddingDType() dtype == dtypes.embedding_dtype True
Bases: DType
1D-sequence dtype with optional axis names.
Attributes:
| Name | Type | Description |
|---|---|---|
x_label |
str
|
Optional name of the x axis. |
y_label |
str
|
Optional name of the y axis. |
Example of usage with defined labels
from renumics.spotlight import dtypes dtype = dtypes.Sequence1DDType("time", "acceleration") str(dtype) 'Sequence1D' dtype.x_label 'time' dtype.y_label 'acceleration'
Example of usage with empty labels
from renumics.spotlight import dtypes dtype = dtypes.Sequence1DDType() dtype == dtypes.sequence_1d_dtype True