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.
extra_validation : bool, optional: If True, additional validation is performed. Throws a ValueError if values of the data are not one of [0, 1], [0, 255], [True, False]. This option affects performance. If False, the mask is converted to dtype np.bool.

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