AlphaMask¶

class AlphaMask(data, origin=None, sly_id=None, class_id=None, labeler_login=None, updated_at=None, created_at=None, extra_validation=True)[source]¶

Bases: Bitmap

Bitmap mask with per-pixel alpha; used for semi-transparent segmentation. Immutable.

AlphaMask geometry for a single Label. AlphaMask object is immutable.

Parameters:

data : np.ndarray¶: AlphaMask mask data. Must be a numpy array with values in range [0, 255].
origin=None¶: PointLocation: top, left corner of AlphaMask. Position of the AlphaMask within image.
sly_id : int, optional¶: AlphaMask ID in Supervisely server.
class_id : int, optional¶: ID of ObjClass to which AlphaMask belongs.
labeler_login : str, optional¶: Login of the user who created AlphaMask.
updated_at : str, optional¶: Date and Time when AlphaMask was modified last. Date Format: Year:Month:Day:Hour:Minute:Seconds. Example: ‘2021-01-22T19:37:50.158Z’.
created_at : str, optional¶: Date and Time when AlphaMask was created. Date Format is the same as in “updated_at” parameter.
extra_validation : bool, optional¶: If True, additional validation is performed. Throws a ValueError if values of the data are not in the range [0, 255]. If True it will affect performance.

:raises ValueError, if data values are not in the range [0, 255].

Usage Example:

import supervisely as sly

# Create simple alpha mask:
mask = np.array([[0, 0, 0, 0, 0],
                [0, 50, 50, 50, 0],
                [0, 50, 0, 50, 0],
                [0, 50, 50, 50, 0],
                [0, 0, 0, 0, 0]], dtype=np.uint8)

figure = sly.AlphaMask(mask)

origin = figure.origin.to_json()
print(json.dumps(origin, indent=4))
# Output: {
#     "points": {
#         "exterior": [
#             [
#                 1,
#                 1
#             ]
#         ],
#         "interior": []
#     }

# Create alpha mask from PNG image:
img = sly.imaging.image.read(os.path.join(os.getcwd(), 'black_white.png'))
mask = img[:, :, 3]
figure = sly.AlphaMask(mask)

Methods

`allowed_transforms`	Returns the allowed transforms for the AlphaMask.
`base64_2_data`	Convert base64 encoded string to numpy array.
`bitwise_mask`	Make bitwise operations between a given numpy array and Bitmap.
`clone`	Clone from GEOMETRYYY
`config_from_json`	Convert geometry config from json format
`config_to_json`	Convert geometry config to json format
`convert`	Convert geometry to another geometry shape.
`crop`	Crops current Bitmap.
`data_2_base64`	Convert numpy array to base64 encoded string.
`draw`
`draw_contour`	Draws the figure contour on a given bitmap canvas :param bitmap: np.ndarray :param color: [R, G, B] :param thickness: (int) :param config: drawing config specific to a concrete subclass, e.g. per edge colors.
`fliplr`	Flip current Bitmap in horizontal.
`flipud`	Flip current Bitmap in vertical.
`from_json`	Convert a json dict to BitmapBase.
`from_path`	Read alpha_channel from image by path.
`geometry_name`	Returns the name of the geometry.
`get_mask`	Returns 2D boolean mask of the geometry.
`name`	Get the name of the geometry.
`relative_crop`	Crops object like "crop" method, but return results with coordinates relative to rect :param rect: Rectangle for crop.
`resize`	Resizes current Bitmap.
`rotate`	Rotates current AlphaMask.
`scale`	Scale current Bitmap.
`skeletonize`	Compute the skeleton, medial axis transform or morphological thinning of Bitmap.
`to_bbox`	Create `Rectangle` object from current Bitmap.
`to_contours`	Get list of contours in Bitmap.
`to_json`	Convert the BitmapBase to a json dict.
`translate`	Translate current Bitmap.
`validate`	Validate geometry.

Attributes

`area`	AlphaMask area.
`data`	Get mask data of Bitmap.
`origin`	Position of the Bitmap within image.

classmethod allowed_transforms()[source]¶: Returns the allowed transforms for the AlphaMask.

classmethod from_json(json_data)¶

Convert a json dict to BitmapBase. Read more about Supervisely format.

Parameters:

json_data : dict¶: Bitmap in json format as a dict.

Returns:

Bitmap.

Return type:

BitmapBase

Usage Example:

import supervisely as sly

figure_json = {
    "bitmap": {
        "origin": [1, 1],
        "data": "eJzrDPBz5+WS4mJgYOD19HAJAtLMIMwIInOeqf8BUmwBPiGuQPr///9Lb86/C2QxlgT5BTM4PLuRBuTwebo4hlTMSa44sKHhISMDuxpTYrr03F6gDIOnq5/LOqeEJgDM5ht6"
    },
    "shape": "bitmap",
    "geometryType": "bitmap"
}

figure = sly.Bitmap.from_json(figure_json)

classmethod from_path(path)[source]¶

Read alpha_channel from image by path.

Parameters:

path : str¶: Path to image

Returns:

Alpha mask

Return type:

AlphaMask

classmethod name()¶

Get the name of the geometry.

Same as geometry_name(), but shorter. In order to make the code more concise.

Returns:: string with name of geometry

static base64_2_data(s)[source]¶

Convert base64 encoded string to numpy array. Supports both compressed and uncompressed masks.

Parameters:

s : str¶: Input base64 encoded string.

Returns:

numpy array

Return type:

np.ndarray

Usage Example:

import supervisely as sly

encoded_string = 'eJzrDPBz5+WS4mJgYOD19HAJAtLMIMwIInOeqf8BUmwBPiGuQPr///9Lb86/C2QxlgT5BTM4PLuRBuTwebo4hlTMSa44sKHhISMDuxpTYrr03F6gDIOnq5/LOqeEJgDM5ht6'
figure_data = sly.AlphaMask.base64_2_data(encoded_string)
print(figure_data)

uncompressed_string = 'iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAARnQU1BAACxjwv8YQUAAAAJcEhZcwAADsMAAA'
mask = sly.AlphaMask.base64_2_data(uncompressed_string)
print(mask)

static config_from_json(config)¶

Convert geometry config from json format

Parameters:

config : dict¶: dictionary(geometry config) in json format

Returns:

dictionary(geometry config) in json format

Return type:

dict

static config_to_json(config)¶

Convert geometry config to json format

Parameters:

config : dict¶: dictionary(geometry config)

Returns:

dictionary(geometry config) in json format

Return type:

dict

static data_2_base64(mask)[source]¶

Convert numpy array to base64 encoded string.

Parameters:

mask : np.ndarray¶: Bool numpy array.

Returns:

Base64 encoded string

Return type:

str

Usage Example:

import os
from dotenv import load_dotenv

import supervisely as sly

# Load secrets and create API object from .env file (recommended)
# Learn more here: https://developer.supervisely.com/getting-started/basics-of-authentication
if sly.is_development():
    load_dotenv(os.path.expanduser("~/supervisely.env"))

api = sly.Api.from_env()

# Get annotation from API
meta_json = api.project.get_meta(PROJECT_ID)
meta = sly.ProjectMeta.from_json(meta_json)
ann_info = api.annotation.download(IMAGE_ID)
ann = sly.Annotation.from_json(ann_info.annotation, meta)

# Get AlphaMask from annotation
for label in ann.labels:
    if type(label.geometry) == sly.AlphaMask:
        figure = label.geometry

encoded_string = sly.AlphaMask.data_2_base64(figure.data)
print(encoded_string)
# 'eJzrDPBz5+WS4mJgYOD19HAJAtLMIMwIInOeqf8BUmwBPiGuQPr///9Lb86/C2QxlgT5BTM4PLuRBuTwebo4hlTMSa44sKHhISMDuxpTYrr03F6gDIOnq5/LOqeEJgDM5ht6'

static geometry_name()[source]¶

Returns the name of the geometry.

Returns:: name of the geometry
Return type:: str

bitwise_mask(full_target_mask, bit_op)¶

Make bitwise operations between a given numpy array and Bitmap.

Parameters:

full_target_mask : np.ndarray¶: Input numpy array.
bit_op¶: Type of bitwise operation(and, or, not, xor), uses numpy logic functions.

Returns:

Bitmap or empty list.

Return type:

Bitmap

Usage Example:

import supervisely as sly

mask = np.array([[0, 0, 0, 0, 0],
                [0, 1, 1, 1, 0],
                [0, 1, 0, 1, 0],
                [0, 1, 1, 1, 0],
                [0, 0, 0, 0, 0]], dtype=np.bool_)

figure = sly.Bitmap(mask)

array = np.array([[0, 0, 0, 0, 0],
                [0, 1, 1, 1, 0],
                [0, 0, 1, 0, 0],
                [0, 0, 1, 0, 0],
                [0, 0, 0, 0, 0]], dtype=np.bool_)

bitwise_figure = figure.bitwise_mask(array, np.logical_and)
print(bitwise_figure.data)
# Output:
# [[ True  True  True]
#  [False False False]
#  [False  True False]]

clone()¶: Clone from GEOMETRYYY

convert(new_geometry, contour_radius=0, approx_epsilon=None)¶

Convert geometry to another geometry shape.

Parameters:

new_geometry¶: New geometry shape.
contour_radius : int¶: Radius of the contour.
approx_epsilon : float¶: Approximation epsilon.

Returns:

List of geometries.

Return type:

List[Geometry]

crop(rect)¶

Crops current Bitmap.

Parameters:

rect¶: Rectangle object for cropping.

Returns:

List of Bitmaps

Return type:

List[Bitmap]

Usage Example:

crop_figures = figure.crop(sly.Rectangle(1, 1, 300, 350))

draw(bitmap, color, thickness=1, config=None)¶

Parameters:

bitmap¶: np.ndarray
color¶: Color [R, G, B]
thickness=1¶: used only in Polyline and Point
config : dict¶: Drawing config specific to a concrete subclass, e.g. per edge colors

Raises:

NotImplementedError – if method is not implemented in subclass

draw_contour(bitmap, color, thickness=1, config=None)¶: Draws the figure contour on a given bitmap canvas :param bitmap: np.ndarray :param color: [R, G, B] :param thickness: (int) :param config: drawing config specific to a concrete subclass, e.g. per edge colors

fliplr(img_size)¶

Flip current Bitmap in horizontal.

Parameters:

img_size : Tuple[int, int]¶: img_size which belongs Bitmap.

Returns:

Flipped bitmap in horizontal.

Return type:

BitmapBase

Usage Example:

# Remember that Bitmap class object is immutable, and we need to assign new instance of Bitmap to a new variable
height, width = 300, 400
fliplr_figure = figure.fliplr((height, width))

flipud(img_size)¶

Flip current Bitmap in vertical.

Parameters:

img_size : Tuple[int, int]¶: img_size which belongs Bitmap.

Returns:

Flipped bitmap in vertical.

Return type:

BitmapBase

Usage Example:

# Remember that Bitmap class object is immutable, and we need to assign new instance of Bitmap to a new variable
height, width = 300, 400
flipud_figure = figure.flipud((height, width))

get_mask(img_size)¶

Returns 2D boolean mask of the geometry. With shape as img_size (height, width) and filled with True values inside the geometry and False values outside. dtype = np.bool shape = img_size

Parameters:

img_size : Tuple[int, int]¶: size of the image (height, width)

Returns:

2D boolean mask of the geometry

Return type:

np.ndarray

relative_crop(rect)¶: Crops object like “crop” method, but return results with coordinates relative to rect :param rect: Rectangle for crop. :type rect: Rectangle :returns: List of Geometry after relative crop. :rtype: List[Geometry]` :raises NotImplementedError: if method is not implemented in subclass

resize(in_size, out_size)¶

Resizes current Bitmap.

Parameters:

in_size : Tuple[int, int]¶: Input image size (height, width) to which Bitmap belongs.
out_size : Tuple[int, int]¶: Output image size (height, width) to which Bitmap belongs.

Returns:

Resized bitmap.

Return type:

Bitmap

Usage Example:

in_height, in_width = 800, 1067
out_height, out_width = 600, 800
# Remember that Bitmap class object is immutable, and we need to assign new instance of Bitmap to a new variable
resize_figure = figure.resize((in_height, in_width), (out_height, out_width))

rotate(rotator)[source]¶

Rotates current AlphaMask.

Parameters:

rotator¶: ImageRotator for AlphaMask rotation.

Returns:

Alpha mask

Return type:

AlphaMask

Usage Example:

import supervisely as sly
from supervisely.geometry.image_rotator import ImageRotator

height, width = ann.img_size
rotator = ImageRotator((height, width), 25)
# Remember that AlphaMask class object is immutable, and we need to assign new instance of AlphaMask to a new variable
rotate_figure = figure.rotate(rotator)

scale(factor)¶

Scale current Bitmap.

Parameters:

factor : float¶: Scale parameter.

Returns:

Scaled bitmap.

Return type:

BitmapBase

Usage Example:

# Remember that Bitmap class object is immutable, and we need to assign new instance of Bitmap to a new variable
scale_figure = figure.scale(0.75)

skeletonize(method_id)¶

Compute the skeleton, medial axis transform or morphological thinning of Bitmap.

Parameters:

method_id¶: Method to convert bool numpy array.

Returns:

Skeletonized bitmap.

Return type:

Bitmap

Usage Example:

# Remember that Bitmap class object is immutable, and we need to assign new instance of Bitmap to a new variable
skeleton_figure = figure.skeletonize(SkeletonizeMethod.SKELETONIZE)
med_ax_figure = figure.skeletonize(SkeletonizeMethod.MEDIAL_AXIS)
thin_figure = figure.skeletonize(SkeletonizeMethod.THINNING)

to_bbox()¶

Create Rectangle object from current Bitmap.

Returns:

Rectangle from bitmap.

Return type:

Rectangle

Usage Example:

rectangle = figure.to_bbox()

to_contours()¶

Get list of contours in Bitmap.

Returns:

List of polygons from bitmap.

Return type:

List[Polygon]

Usage Example:

figure_contours = figure.to_contours()

to_json()¶

Convert the BitmapBase to a json dict. Read more about Supervisely format.

Returns:

Json format as a dict

Return type:

dict

Usage Example:

import supervisely as sly

mask = np.array([[0, 0, 0, 0, 0],
                [0, 1, 1, 1, 0],
                [0, 1, 0, 1, 0],
                [0, 1, 1, 1, 0],
                [0, 0, 0, 0, 0]], dtype=np.bool_)

figure = sly.Bitmap(mask)
figure_json = figure.to_json()
print(json.dumps(figure_json, indent=4))
# Output: {
#    "bitmap": {
#        "origin": [1, 1],
#        "data": "eJzrDPBz5+WS4mJgYOD19HAJAtLMIMwIInOeqf8BUmwBPiGuQPr///9Lb86/C2QxlgT5BTM4PLuRBuTwebo4hlTMSa44sKHhISMDuxpTYrr03F6gDIOnq5/LOqeEJgDM5ht6"
#    },
#    "shape": "bitmap",
#    "geometryType": "bitmap"
# }

translate(drow, dcol)¶

Translate current Bitmap.

Parameters:

drow : int¶: Horizontal shift.
dcol : int¶: Vertical shift.

Returns:

Translated bitmap.

Return type:

BitmapBase

Usage Example:

# Remember that Bitmap class object is immutable, and we need to assign new instance of Bitmap to a new variable
translate_figure = figure.translate(150, 250)

validate(obj_class_shape, settings)¶

Validate geometry.

Parameters:

obj_class_shape : str¶: Object class shape.
settings : dict¶: Settings.

Raises:

ValueError – if geometry validation error

property area : float¶

AlphaMask area.

Returns:

Area of current AlphaMask

Return type:

float

Usage Example:

print(figure.area)
# Output: 88101.0

property data : numpy.ndarray¶

Get mask data of Bitmap.

Returns:: Data of Bitmap.
Return type:: np.ndarray

property origin : supervisely.geometry.point_location.PointLocation¶

Position of the Bitmap within image.

Returns:: Top, left corner of Bitmap.
Return type:: PointLocation