Bitmap¶

Bases: supervisely.geometry.bitmap_base.BitmapBase

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

Parameters

data : np.ndarray: Bitmap mask data. Must be a numpy array with only 2 unique values: [0, 1] or [0, 255] or [False, True].
origin : PointLocation, optional: PointLocation: top, left corner of Bitmap. Position of the Bitmap within image.
sly_id : int, optional: Bitmap ID in Supervisely server.
class_id : int, optional: ID of ObjClass to which Bitmap belongs.
labeler_login : str, optional: Login of the user who created Bitmap.
updated_at : str, optional: Date and Time when Bitmap 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 Bitmap was created. Date Format is the same as in “updated_at” parameter.

Raises

ValueError, if data is not bool or no pixels set to True in data

Usage example

import supervisely as sly

# Create simple bitmap
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)

# Note, when creating a bitmap, the specified mask is cut off by positive values, based on this, a origin is formed:
print(figure.data)
# Output:
#    [[ True  True  True]
#     [ True False  True]
#     [ True  True  True]]

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

# Create bitmap from black and white image:
img = sly.imaging.image.read(os.path.join(os.getcwd(), 'black_white.jpeg'))
mask = img[:, :, 0].astype(bool) # Get 2-dimensional bool numpy array
figure = sly.Bitmap(mask)

Methods

`allowed_transforms`
`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`
`config_to_json`
`convert`
`crop`	Crops current Bitmap.
`data_2_base64`	Convert numpy array to base64 encoded string.
`draw`	param bitmap np.ndarray
`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.
`fliplr`	Flip current Bitmap in horizontal.
`flipud`	Flip current Bitmap in vertical.
`from_json`	Convert a json dict to BitmapBase.
`from_path`	Read bitmap from image by path.
`geometry_name`
`get_mask`	Returns 2D boolean mask of the geometry.
`name`	Same as geometry_name(), but shorter.
`relative_crop`	Crops object like "crop" method, but return results with coordinates relative to rect :param rect: :return: list of Geometry
`resize`	Resizes current Bitmap.
`rotate`	Rotates current Bitmap.
`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`

Attributes

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

classmethod allowed_transforms()[source]¶

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

Bool numpy array

Return type

np.ndarray

Usage example

import supervisely as sly

encoded_string = 'eJzrDPBz5+WS4mJgYOD19HAJAtLMIMwIInOeqf8BUmwBPiGuQPr///9Lb86/C2QxlgT5BTM4PLuRBuTwebo4hlTMSa44sKHhISMDuxpTYrr03F6gDIOnq5/LOqeEJgDM5ht6'
figure_data = sly.Bitmap.base64_2_data(encoded_string)
print(figure_data)
#  [[ True  True  True]
#   [ True False  True]
#   [ True  True  True]]

uncompressed_string = 'iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAAAAXNSR0IArs4c6QAAAARnQU1BAACxjwv8YQUAAAAJcEhZcwAADsMAAA'
boolean_mask = sly.Bitmap.base64_2_data(uncompressed_string)
print(boolean_mask)
#  [[ True  True  True]
#   [ True False  True]
#   [ True  True  True]]

bitwise_mask(full_target_mask, bit_op)[source]¶

Make bitwise operations between a given numpy array and Bitmap.

Parameters

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

Returns

Bitmap object or empty list

Return type

Bitmap or list

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

crop(rect)[source]¶

Crops current Bitmap.

Parameters

rect : Rectangle: Rectangle object for cropping.

Returns

List of Bitmaps

Return type

List[Bitmap]

Usage Example

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

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 supervisely as sly

address = 'https://app.supervise.ly/'
token = 'Your Supervisely API Token'
api = sly.Api(address, token)

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

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

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

Parameters

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

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]: Annotation.img_size which belongs Bitmap.

Returns

BitmapBase object

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]: Annotation.img_size which belongs Bitmap.

Returns

BitmapBase object

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))

classmethod from_json(json_data)¶

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

Parameters

data : dict: Bitmap in json format as a dict.

Returns

BitmapBase object

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 bitmap from image by path.

Parameters

path : str: Path to image

Returns

Bitmap

Return type

Bitmap

static geometry_name()[source]¶

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

classmethod name()¶

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

Returns: string with name of geometry

relative_crop(rect)¶: Crops object like “crop” method, but return results with coordinates relative to rect :param rect: :return: list of Geometry

resize(in_size, out_size)[source]¶

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

Bitmap object

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 Bitmap.

Parameters

rotator : ImageRotator: ImageRotator for Bitamp rotation.

Returns

Bitmap object

Return type

Bitmap

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 Bitmap class object is immutable, and we need to assign new instance of Bitmap to a new variable
rotate_figure = figure.rotate(rotator)

scale(factor)¶

Scale current Bitmap.

Parameters

factor : float: Scale parameter.

Returns

BitmapBase object

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)[source]¶

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

Parameters

method_id : SkeletonizeMethod: Method to convert bool numpy array.

Returns

Bitmap object

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 object

Return type

Rectangle

Usage Example

rectangle = figure.to_bbox()

to_contours()[source]¶

Get list of contours in Bitmap.

Returns

List of Polygon objects

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

BitmapBase object

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)

property area¶

Bitmap area.

Returns

Area of current Bitmap

Return type

float

Usage example

print(figure.area)
# Output: 88101.0

property data¶

Get mask data of Bitmap.

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

property origin¶

Position of the Bitmap within image.

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