Features Subpackage

cv2_group.features.feature_extraction.analyze_primary_root(root_instances: List[ndarray | None], original_roi_indices: List[int]) → List[Dict[str, Any]]

Analyze each primary root binary mask instance to extract geometric and path features.

Parameters:

  • root_instances (list of np.ndarray or None) – List of binary masks representing individual root instances. Elements can be None if no root was detected.

  • original_roi_indices (list of int) – Indices mapping each root instance back to the original ROI.

Returns:

Each dict contains:

  • ’roi_index’ (int): ROI index for this root.

  • ’length’ (float): Length of the primary root path.

  • ’tip_coords’ (list or None): Coordinates [y, x] of the root tip.

  • ’base_coords’ (list or None): Coordinates [y, x] of the root base.

  • ’primary_path’ (list or None): List of coordinates of the primary root.

Return type:

list of dict

cv2_group.features.feature_extraction.extract_root_instances(label_ids: ndarray, max_label_ids_for_rois: List[int | None]) → List[ndarray | None]

Extract binary masks for individual root instances from a labeled image. Each extracted mask will have values 0 or 1.

Parameters:

  • label_ids (np.ndarray) – Array with labeled connected components.

  • max_label_ids_for_rois (List[Optional[int]]) – List of label IDs to extract, corresponding to identified ROIs. Can contain None for ROIs where no label was found.

Returns:

List of binary masks (0 or 1) for each root instance, or None if no instance was found for a given ROI.

Return type:

List[Optional[np.ndarray]]

cv2_group.features.feature_extraction.find_labels_in_rois(label_ids: ndarray, total_labels: int, stats: ndarray, rois: List[Tuple[int, int, int, int]]) → Tuple[Dict[str, int | None], List[int]]

Identify root labels within defined regions of interest (ROIs). Prioritizes the label with the maximum height within each ROI. Handles cases where no ROIs are provided, returning no specific labels.

Parameters:

  • label_ids (np.ndarray) – Array with labeled connected components.

  • total_labels (int) – Total number of labels found.

  • stats (np.ndarray) – Statistics for each connected component.

  • rois (List[Tuple[int, int, int, int]]) – List of ROIs, each as (x, y, width, height).

Returns:

  • max_label_dict: Dictionary mapping “label_idx” to the identified

    label ID, or None if no label found in ROI.

  • max_label_list: List of identified label IDs (can contain None).

Return type:

Tuple[Dict[str, Optional[int]], List[int]]

cv2_group.features.feature_extraction.process_predicted_mask(predicted_mask: ndarray) → Tuple[ndarray, ndarray, int, ndarray, ndarray]

Process a binary mask to generate labeled root instances using connected components analysis. This function is designed to accept masks from both model prediction and user edits.

Parameters:

predicted_mask (np.ndarray) – Input mask (2D array), values can be 0 or 1, or 0-255. It will be normalized to 0 or 255 (uint8) for processing.

Returns:

  • binary_mask: The input mask ensured to be 0/255 uint8.

  • label_ids: Labeled connected components.

  • total_labels: Total number of labels found.

  • stats: Statistics for each connected component.

  • centroids: Centroids for each connected component.

Return type:

Tuple[np.ndarray, np.ndarray, int, np.ndarray, np.ndarray]