Detection Result Category Selection

Description

This operator is used to filter detection results of a specific category from detection instances, facilitating more precise subsequent target analysis.

Overview

Filters detection instances by class, with optional fallback to the original results.

Scenarios

  • Quickly extracting targets of a specific category in multi-category detection scenarios.

  • Subsequent analysis processes target only a single object category, such as target statistics, tracking, or classification tasks for a specific category.

  • When the target category is not detected, you can choose whether to use detection instances of other categories as an alternative.

Inputs and Outputs

Inputs

Detection Results: Detection instances output by deep learning models or other detection algorithms, including the bounding box, classification information, confidence score, and polygon contour of each target.

Outputs

Single category detection results: Detection instances containing only the specified category after filtering.

Parameters

Category

Description

Selects the class number to retain from detection results.

Tuning Instructions

When the node is enabled, only instances whose class number matches the selection are returned.

Parameter Range

Options: 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, Default: 0

Parameter Range

Values/Range:

  • 0

  • 1

  • 2

  • 3

  • 4

  • 5

  • 6

  • 7

  • 8

  • 9

Default: 0

Use Substitute Category

Description

Controls whether other classes are returned when the selected class has no results.

Tuning Instructions

Used only when the node is enabled and the selected class is absent; enabling it returns the original results.

Parameter Range

Default: False

Enable Node

Description

Controls whether class filtering is applied.

Tuning Instructions

Disabling it bypasses filtering and returns the input detection results unchanged.

Parameter Range

Default: True