draw.instance_mask

Description

Draws instance segmentation masks.

class Node(config=None, **kwargs)

Draws instance segmentation masks on image.

The draw.mask node draws instance segmentation masks onto the detected object instances.

Inputs

img (numpy.ndarray): A NumPy array of shape \((height, width, channels)\) containing the image data in BGR format.

masks (numpy.ndarray): A NumPy array of shape \((N, H, W)\) containing \(N\) detected binarized masks where \(H\) and \(W\) are the height and width of the masks. The order corresponds to bbox_labels.

bbox_labels (numpy.ndarray): A NumPy array of shape \((N)\) containing strings representing the labels of detected objects. The order corresponds to bboxes and bbox_scores.

Outputs

img (numpy.ndarray): A NumPy array of shape \((height, width, channels)\) containing the image data in BGR format.

Configs

• instance_color_scheme (str) – {“random”, “hue_family”}, default = “hue_family”
  This defines what colors to use for the standard masks.
  “hue_family”: use the same hue for each instance belonging to the same class, but with a slightly different saturation.
  “random”: use a random color for all instances.

• effect (dict) – {contrast: null, brightness: null, gamma_correction: null, blur: null, mosaic: null}
  This defines the effect (if any) to apply to either the masked (objects) or unmasked (background) areas of the image. If no effect is selected, a “standard” instance segmentation mask will be drawn and colored according to the instance_color_scheme.
  For example, to apply the contrast effect to the objects in the image, set the following config in pipeline_config.yml:

  effect : {contrast: 1.2}

  Note that at most one effect can be enabled at a time.

  Effect	Description	Data Type	Range
  contrast	Adjusts contrast using this value as the “alpha” parameter.	float	[0.0, 3.0]
  brightness	Adjusts brightness using this value as the “beta” parameter.	int	[-100, 100]
  gamma_correction	Adjusts gamma using this value as the “gamma” parameter.	float	[0.0, +inf]
  blur	Blurs the area using this value as the “blur_kernel_size” parameter. Larger values gives more intense blurring.	int	[1, +inf]
  mosaic	Mosaics the area using this value as the resolution of a mosaic filter (width × height). The number corresponds to the number of rows and columns used to create a mosaic. For example, the setting (mosaic: 25) creates a \(25 \times 25\) mosaic filter. Increasing the number increases the intensity of pixelation over an area.	int	[1, +inf]

• effect_area (str) – {“objects”, “background”}, default = “objects”
  This defines where the effect should be applied.
  “objects”: the effect is applied to the masked areas of the image.
  “background”: the effect is applied to the unmasked areas of the image.

• contours (dict) – {show: False, thickness: 2}
  

  Contours	Description	Data Type	Range
  show	This determines whether to show the contours of the masks.	bool	N.A.
  thickness	This defines the thickness of the contours.	int	[1, +inf]