Extract Attributes

Overview

Extracts a specific attribute from prediction data produced by a model node. This node allows you to isolate individual components such as polygon coordinates, confidence scores, or classification results for further processing.

Use Extract Attributes when:

• You want to access only the confidence scores (probability) from a model's prediction.
• You need to filter polygons based on their class or area.

💡 Tip

This node is essential when working with model outputs set to 'mapped' mode, allowing you to access specific data fields within the prediction.

Inputs

Name	Type	Description
input	any	The raw prediction data from a model node.

Outputs

Name	Type	Description
extracted	dynamic	The specific attribute value extracted from the input, type depends on chosen parameter.

Parameters

Parameter	Default	Valid values	Tunable
Name of attribute to extract	polygons	polygons, probability, prediction_class, points	❌

Name of attribute to extract

The specific field to extract from the input data.

Example

Extract confidence scores

A model outputs complex prediction objects that contain polygons, probability or points based problem type. You need to access the probabilities to decide if a detection is reliable.

Configuration:

• Name of attribute to extract: probability

Result: The node outputs the list of confidence scores for all detected objects.

Related links

Reference:

• AI Model — standard node for running AI inferences
• Polygon Threshold — filters polygons based on criteria