Deepskin segmentation

deepskin.segmentation.wound_segmentation(img: ndarray, tol: float = 0.5, verbose: bool = False) → ndarray[source]

Perform the semantic image segmentation using the Deepskin semantic model.

Parameters:

  • img (np.ndarray) – Input image to analyze in RGB format

  • tol (float (default := 0.5)) – Threshold to apply on the resulting mask for the output binarization

  • verbose (bool (default := False)) – Enable/Disable the logging of the steps

Returns:

pred – Output image mask obtained by the model, in which the semantic meaning is organized as: background (channel 0), body (channel 1), wound (channel 2)

Return type:

np.ndarray