fruit_project.utils.metrics
===========================

.. py:module:: fruit_project.utils.metrics


Classes
-------

.. autoapisummary::

   fruit_project.utils.metrics.ConfusionMatrix
   fruit_project.utils.metrics.MAPEvaluator


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

.. py:class:: ConfusionMatrix(nc: int, conf: float = 0.25, iou_thres: float = 0.45)

   Object Detection Confusion Matrix inspired by Ultralytics.

   :param nc: Number of classes.
   :type nc: int
   :param conf: Confidence threshold for detections.
   :type conf: float
   :param iou_thres: IoU threshold for matching.
   :type iou_thres: float


   .. py:attribute:: nc


   .. py:attribute:: conf
      :value: 0.25



   .. py:attribute:: iou_thres
      :value: 0.45



   .. py:attribute:: matrix


   .. py:attribute:: eps
      :value: 1e-06



   .. py:method:: process_batch(detections: torch.Tensor, labels: torch.Tensor) -> None

      Update the confusion matrix with a batch of detections and ground truths.

      :param detections: Tensor of detections, shape [N, 6] (x1, y1, x2, y2, conf, class).
      :type detections: torch.Tensor
      :param labels: Tensor of ground truths, shape [M, 5] (class, x1, y1, x2, y2).
      :type labels: torch.Tensor



   .. py:method:: update(preds, targets_for_cm)


   .. py:method:: plot(class_names: List[str], normalize: bool = True) -> matplotlib.pyplot.Figure

      Generates and returns a matplotlib figure of the confusion matrix.



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

      Returns the raw confusion matrix tensor.

      :returns: The (nc + 1) x (nc + 1) confusion matrix.
      :rtype: torch.Tensor



.. py:class:: MAPEvaluator(image_processor, device, threshold: float = 0.0, id2label: Optional[Dict[int, str]] = None)

   Mean Average Precision evaluator for RT-DETRv2 - adapted for fruit_project.


   .. py:attribute:: image_processor


   .. py:attribute:: threshold
      :value: 0.0



   .. py:attribute:: id2label
      :value: None



   .. py:attribute:: map_metric


   .. py:attribute:: map_50_metric


   .. py:attribute:: device


   .. py:method:: collect_image_sizes(targets)

      Collect image sizes from targets.



   .. py:method:: collect_targets(targets, image_sizes)

      Process ground truth targets - now handles HF-processed format.



   .. py:method:: collect_predictions(predictions, image_sizes)

      Process model predictions using HuggingFace post-processing.



   .. py:method:: get_per_class(map_50_metrics, metric)


   .. py:method:: get_optimal_f1_ultralytics_style(metrics_dict)


   .. py:method:: get_averaged_precision_recall_ultralytics_style(optimal_precisions: torch.Tensor, optimal_recalls: torch.Tensor, present_classes: torch.Tensor)

      Calculate overall precision and recall...