def make_binary_mask(probs: np.ndarray, threshold: float = -1) -> np.ndarray:
"""
Computes the binary mask of the detected class from probabilities
:param probs: array with values in range [0, 1]
:param threshold: threshold between [0 and 1], if negative Otsu's adaptive threshold will be used
:return: binary mask
"""
mask = binarization.thresholding(probs, threshold)
mask = binarization.cleaning_binary(mask, kernel_size=5)
return mask
def post_process_probs_ornament(probability_maps):
binary_maps = np.zeros_like(probability_maps, np.uint8)
binary_maps = np.delete(binary_maps, 0, 2)
# Ornament
binary_image = binarization.thresholding(probability_maps[:, :, 1],
threshold=0.75)
binary_image = binarization.cleaning_binary(binary_image, kernel_size=3)
boxes = boxes_detection.find_boxes(binary_image,
mode='rectangle',
min_area=0.)
bin_map = np.zeros_like(binary_maps)
binary_maps[:, :, 0] = cv2.fillPoly(bin_map, boxes, (255, 0, 0))[:, :, 0]
return binary_maps, boxes
def page_post_processing_fn(probs: np.ndarray, threshold: float=0.5, output_basename: str=None,
kernel_size: int = 5) -> np.ndarray:
"""
Computes the binary mask of the detected Page from the probabilities outputed by network
:param probs: array in range [0, 1] of shape HxWx2
:param threshold: threshold between [0 and 1], if negative Otsu's adaptive threshold will be used
:param output_basename:
:param kernel_size: size of kernel for morphological cleaning
"""
mask = binarization.thresholding(probs[:, :, 1], threshold=threshold)
result = binarization.cleaning_binary(mask, size=kernel_size)
if output_basename is not None:
imsave('{}.png'.format(output_basename), result*255)
return result