Polygon¶

class Polygon[source]¶

Bases: supervisely.geometry.vector_geometry.VectorGeometry

Polygon geometry for a single Label. Polygon class object is immutable.

Parameters

exterior : List[PointLocation], List[List[int, int]], List[Tuple[int, int]: Exterior coordinates, object contour is defined with these points.
interior : List[List[PointLocation]], List[List[List[int, int]]], List[List[Tuple[int, int]]]: Interior coordinates, object holes are defined with these points.
sly_id : int, optional: Polygon ID in Supervisely server.
class_id : int, optional: ID of ObjClass to which Polygon belongs.
labeler_login : str, optional: Login of the user who created Polygon.
updated_at : str, optional: Date and Time when Polygon 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 Polygon was created. Date Format is the same as in “updated_at” parameter.

Raises

ValueError, if len(exterior) < 3 or len(any element in interior list) < 3

Usage example

import supervisely as sly

exterior = [sly.PointLocation(730, 2104), sly.PointLocation(2479, 402), sly.PointLocation(3746, 1646)]
# or exterior = [[730, 2104], [2479, 402], [3746, 1646]]
# or exterior = [(730, 2104), (2479, 402), (3746, 1646)]
interior = [[sly.PointLocation(1907, 1255), sly.PointLocation(2468, 875), sly.PointLocation(2679, 1577)]]
# or interior = [[[730, 2104], [2479, 402], [3746, 1646]]]
# or interior = [[(730, 2104), (2479, 402), (3746, 1646)]]
figure = sly.Polygon(exterior, interior)

Methods

`allowed_transforms`
`approx_dp`	Approximates a Polygon curve with the specified precision.
`clone`	Clone from GEOMETRYYY
`config_from_json`
`config_to_json`
`convert`
`crop`	Crops current Polygon.
`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`	Flips current VectorGeometry in horizontal.
`flipud`	Flips current VectorGeometry in vertical.
`from_json`	Convert a json dict to Polygon.
`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 VectorGeometry.
`rotate`	Rotates current VectorGeometry.
`scale`	Scales current VectorGeometry.
`to_bbox`	Creates Rectangle object from current VectorGeometry.
`to_json`	Convert the VectorGeometry to a json dict.
`translate`	Translates current VectorGeometry.
`validate`

Attributes

`area`	Polygon area.
`exterior`	VectorGeometry exterior points.
`exterior_np`	Converts exterior attribute of VectorGeometry to numpy array.
`interior`	VectorGeometry interior points.
`interior_np`	Converts interior attribute of VectorGeometry to numpy array.

classmethod allowed_transforms()[source]¶

approx_dp(epsilon)[source]¶

Approximates a Polygon curve with the specified precision.

Parameters

epsilon : float: Specifying the approximation accuracy.

Returns

Polygon object

Return type

Polygon

Usage Example

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

clone()¶: Clone from GEOMETRYYY

crop(rect)[source]¶

Crops current Polygon.

Parameters

rect : Rectangle: Rectangle object for crop.

Returns

List of Polygon objects

Return type

List[Polygon]

Usage Example

import supervisely as sly

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

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)¶

Flips current VectorGeometry in horizontal.

Parameters

img_size : Tuple[int, int]: Input image size (height, width) to which belongs VectorGeometry.

Returns

VectorGeometry object

Return type

VectorGeometry

Usage Example

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

flipud(img_size)¶

Flips current VectorGeometry in vertical.

Parameters

img_size : Tuple[int, int]: Input image size (height, width) to which belongs VectorGeometry.

Returns

VectorGeometry object

Return type

VectorGeometry

Usage Example

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

classmethod from_json(data)[source]¶

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

Parameters

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

Returns

Polygon object

Return type

Polygon

Usage example

import supervisely as sly

figure_json =  {
    "points": {
        "exterior": [
            [2104, 730],
            [402, 2479],
            [1646, 3746]
        ],
        "interior": [
            [
                [1255, 1907],
                [875, 2468],
                [577, 2679]
            ]
        ]
    }
}

figure = sly.Polygon.from_json(figure_json)

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)¶

Resizes current VectorGeometry.

Parameters

in_size : Tuple[int, int]: Input image size (height, width) to which belongs VectorGeometry.
out_size : Tuple[int, int]: Desired output image size (height, width) to which belongs VectorGeometry.

Returns

VectorGeometry object

Return type

VectorGeometry

Usage Example

# Remember that VectorGeometry class object is immutable, and we need to assign new instance of VectorGeometry to a new variable
in_height, in_width = 300, 400
out_height, out_width = 600, 800
resize_figure = figure.resize((in_height, in_width), (out_height, out_width))

rotate(rotator)¶

Rotates current VectorGeometry.

Parameters

rotator : ImageRotator: ImageRotator object for rotate.

Returns

VectorGeometry object

Return type

VectorGeometry

Usage Example

from supervisely.geometry.image_rotator import ImageRotator

# Remember that VectorGeometry class object is immutable, and we need to assign new instance of VectorGeometry to a new variable
height, width = 300, 400
rotator = ImageRotator((height, width), 25)
rotate_figure = figure.rotate(rotator)

scale(factor)¶

Scales current VectorGeometry.

Parameters

factor : float: Scale parameter.

Returns

VectorGeometry object

Return type

VectorGeometry

Usage Example

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

to_bbox()¶

Creates Rectangle object from current VectorGeometry.

Returns

Rectangle object

Return type

Rectangle

Usage Example

rectangle = figure.to_bbox()

to_json()¶

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

Returns

Json format as a dict

Return type

dict

Usage example

figure_json = figure.to_json()
print(figure_json)
# Output: {
#    "points": {
#        "exterior": [
#            [2104, 730],
#            [402, 2479],
#            [1646, 3746]
#        ],
#        "interior": [
#            [
#                [1255, 1907],
#                [875, 2468],
#                [1577, 2679]
#            ]
#        ]
#    }
# }

translate(drow, dcol)¶

Translates current VectorGeometry.

Parameters

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

Returns

VectorGeometry object

Return type

VectorGeometry

Usage Example

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

property area¶

Polygon area.

Returns

Area of current Polygon object (exterior figure only).

Return type

float

Usage Example

print(figure.area)
# Output: 7288.0

property exterior¶

VectorGeometry exterior points.

Returns

VectorGeometry exterior points

Return type

List[PointLocation]

Usage example

exterior = figure.exterior

property exterior_np¶

Converts exterior attribute of VectorGeometry to numpy array.

Returns

Numpy array

Return type

np.ndarray

Usage example

print(figure.exterior_np)
# Output:
# [[ 730 2104]
#  [2479  402]
#  [3746 1646]]

property interior¶

VectorGeometry interior points.

Returns

VectorGeometry interior points

Return type

List[List[PointLocation]]

Usage example

interior = figure.interior

property interior_np¶

Converts interior attribute of VectorGeometry to numpy array.

Returns

Numpy array

Return type

List[np.ndarray]

Usage example

print(figure.interior_np)
# Output:
# [array([[1907, 1255],
#        [2468,  875],
#        [2679, 1577]])]