Skip to content

Detections List Roll-Up

Class: DetectionsListRollUpBlockV1

Source: inference.core.workflows.core_steps.fusion.detections_list_rollup.v1.DetectionsListRollUpBlockV1

Rolls up dimensionality from children to parent detections

Useful in scenarios like: * rolling up results from a secondary model run on crops back to parent images * rolling up OCR results for dynamically cropped images

Type identifier

Use the following identifier in step "type" field: roboflow_core/detections_list_rollup@v1to add the block as as step in your workflow.

Properties

Name Type Description Refs
name str Enter a unique identifier for this step.. ❌
confidence_strategy str Strategy to use when merging confidence scores from child detections. Options are 'max', 'mean', or 'min'.. ✅
overlap_threshold float Minimum overlap ratio (IoU) to consider when merging overlapping detections from child crops. A value of 0.0 merges any overlapping detections, while higher values require greater overlap to merge. Specify between 0.0 and 1.0. A value of 1.0 only merges completely overlapping detections.. ✅
keypoint_merge_threshold float Keypoint distance (in pixels) to merge keypoint detections if the child detections contain keypoint data.. ✅

The Refs column marks possibility to parametrise the property with dynamic values available in workflow runtime. See Bindings for more info.

Available Connections

Compatible Blocks

Check what blocks you can connect to Detections List Roll-Up in version v1.

Input and Output Bindings

The available connections depend on its binding kinds. Check what binding kinds Detections List Roll-Up in version v1 has.

Bindings

  • input

    • parent_detection (Union[instance_segmentation_prediction, keypoint_detection_prediction, object_detection_prediction]): The parent detection the dimensionality inherits from..
    • child_detections (list_of_values): A list of child detections resulting from higher dimensionality, such as predictions made on dynamic crops. Use the "Dimension Collapse" to reduce the higher dimensionality result to one that can be used with this. Example: Prediction -> Dimension Collapse -> Detections List Roll-Up.
    • confidence_strategy (list_of_values): Strategy to use when merging confidence scores from child detections. Options are 'max', 'mean', or 'min'..
    • overlap_threshold (float_zero_to_one): Minimum overlap ratio (IoU) to consider when merging overlapping detections from child crops. A value of 0.0 merges any overlapping detections, while higher values require greater overlap to merge. Specify between 0.0 and 1.0. A value of 1.0 only merges completely overlapping detections..
    • keypoint_merge_threshold (float): Keypoint distance (in pixels) to merge keypoint detections if the child detections contain keypoint data..

  • output

    • rolled_up_detections (Union[instance_segmentation_prediction, object_detection_prediction, keypoint_detection_prediction]): Prediction with detected bounding boxes and segmentation masks in form of sv.Detections(...) object if instance_segmentation_prediction or Prediction with detected bounding boxes in form of sv.Detections(...) object if object_detection_prediction or Prediction with detected bounding boxes and detected keypoints in form of sv.Detections(...) object if keypoint_detection_prediction.
    • crop_zones (list_of_values): List of values of any type.
Example JSON definition of step Detections List Roll-Up in version v1 
{
    "name": "<your_step_name_here>",
    "type": "roboflow_core/detections_list_rollup@v1",
    "parent_detection": "<block_does_not_provide_example>",
    "child_detections": "<block_does_not_provide_example>",
    "confidence_strategy": "min",
    "overlap_threshold": 0.0,
    "keypoint_merge_threshold": 0.0
}