General Object Detection

Overview

General Object Detection node provides ready-to-use object detection with a built-in model and class filtering.

Compared with Object Detection, this node is optimized for quick setup: pick a model size, list the labels you care about, and tune confidence/NMS only if needed. No external model files are required.

Input

Input Image

image required

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

Model size

string required

Size of the built-in detection model. Larger models are more accurate but slower.

Values:

• TINY — fastest; suited for high-frame-rate flows or resource-limited hardware.
• SMALL (default) — balanced speed and accuracy.

Label names to detect

array required

List of label names to detect. Detections for labels not in this list are discarded.

Supported labels include: person, bicycle, car, motorbike, bus, truck, cat, dog, bird, horse, sheep, cow, elephant, bear, zebra, giraffe, knife, phone, scissors.

Default: person, car

Confidence threshold

number required advanced

Minimum confidence score to keep a detection. See Confidence Threshold for tuning guidance.

Range: 0.0 to 1.0
Default: 0.5

Overlap threshold

number required advanced

NMS overlap threshold (higher allows more overlap). See Overlap threshold for tuning guidance.

Range: 0.0 to 1.0
Default: 0.3

Class-agnostic NMS

boolean advanced

When enabled, NMS is applied across all classes. When disabled, NMS runs per class.

Default: true

Use TensorRT if available

boolean required advanced

Enable TensorRT inference backend when available. Can improve throughput on NVIDIA GPUs.

Default: false

Overlay results

boolean required advanced

Whether to draw detections on the output frame. See Overlay Results.

Default: true

Output

Overlay Image

image

Frame with detected object overlays.

Detected Count

integer

Number of detected objects.

Detected Objects

array

Array of detection objects with:

• Bounding Box (bbox): [x, y, width, height]
• Label
• Confidence