Auto Label Object Detection

Overview

Auto Label Object Detection node exports an input image and its rectangular detections as labeled training data in YOLO or Pascal VOC format.

Use this node after a detector such as Object Detection or General Object Detection to capture auto-labeled samples for model training or review.

Input

Input Image

image required

The image frame to export. Connect this to a camera or upstream image output.

Rectangular Objects

array required

List of rectangular detection objects to export. Each object must include at least a Bounding Box (bbox: [x, y, width, height]). Label uses a default when missing.

Export directory

string required

Path to the folder where labeled images and annotation files are written.

Label type

string required

Export annotation format.

Values:

• YOLO — YOLO
• PASCAL_VOC (default) — Pascal VOC

Class name override

array

Optional mappings from the current class name to an override name used in the exported labels. For reference tools that do not provide a class name, use an empty current class name.

Each entry:

• Class name current (default "")
• Class name override (default detected)

Default: []

Filename

string

Filename for the exported labels and images. Supports template expressions {{}}.

Default: ""

Add filename prefix

boolean

Whether to add an auto-generated prefix to the filename to ensure uniqueness.

Default: true

Image file format

string required

Export image file format and extension.

Values:

• JPEG (default)
• JPG
• PNG
• TIFF

Image compression

integer required

Export image compression percentage. Higher values produce more compressed images.

Range: 0 to 100
Default: 5

Output

This node writes labeled images and annotation files to the export directory. It does not expose additional image or detection ports beyond common tool status.