:py:mod:`anomalib.utils.metrics.pro`
====================================

.. py:module:: anomalib.utils.metrics.pro

.. autoapi-nested-parse::

   Implementation of PRO metric based on TorchMetrics.



Module Contents
---------------

Classes
~~~~~~~

.. autoapisummary::

   anomalib.utils.metrics.pro.PRO



Functions
~~~~~~~~~

.. autoapisummary::

   anomalib.utils.metrics.pro.pro_score
   anomalib.utils.metrics.pro.connected_components_gpu
   anomalib.utils.metrics.pro.connected_components_cpu



.. py:class:: PRO(threshold: float = 0.5, **kwargs)

   Bases: :py:obj:`torchmetrics.Metric`

   Per-Region Overlap (PRO) Score.

   .. py:attribute:: target
      :annotation: :List[torch.Tensor]

      

   .. py:attribute:: preds
      :annotation: :List[torch.Tensor]

      

   .. py:method:: update(predictions: torch.Tensor, targets: torch.Tensor) -> None

      Compute the PRO score for the current batch.


   .. py:method:: compute() -> torch.Tensor

      Compute the macro average of the PRO score across all regions in all batches.



.. py:function:: pro_score(predictions: torch.Tensor, comps: torch.Tensor, threshold: float = 0.5) -> torch.Tensor

   Calculate the PRO score for a batch of predictions.

   :param predictions: Predicted anomaly masks (Bx1xHxW)
   :type predictions: Tensor
   :param comps: (Tensor): Labeled connected components (BxHxW). The components should be labeled from 0 to N
   :param threshold: When predictions are passed as float, the threshold is used to binarize the predictions.
   :type threshold: float

   :returns: Scalar value representing the average PRO score for the input batch.
   :rtype: Tensor


.. py:function:: connected_components_gpu(binary_input: torch.Tensor, num_iterations: int = 1000) -> torch.Tensor

   Perform connected component labeling on GPU and remap the labels from 0 to N.

   :param binary_input: Binary input data from which we want to extract connected components (Bx1xHxW)
   :type binary_input: Tensor
   :param num_iterations: Number of iterations used in the connected component computation.
   :type num_iterations: int

   :returns: Components labeled from 0 to N.
   :rtype: Tensor


.. py:function:: connected_components_cpu(image: torch.Tensor) -> torch.Tensor

   Connected component labeling on CPU.

   :param image: Binary input data from which we want to extract connected components (Bx1xHxW)
   :type image: Tensor

   :returns: Components labeled from 0 to N.
   :rtype: Tensor