Object Detection and Counting

R2026b

Detect and count objects in images for visual inspection applications using deep learning
Since R2026b

Detect, localize, and count objects in images for visual inspection applications. Visual Inspection Toolbox™ provides capabilities for these tasks:

Object detection — Train a You Only Look Once X (YOLOX) network to detect and localize objects with bounding boxes. Detect small objects in large images in a single pass by using tiled training and full-resolution inference, without external tiling or stitching. Generate annotated data sets and create synthetic training data to build robust detectors. Export trained models to the Open Neural Network Exchange (ONNX) format for deployment.
Object counting — Use exemplar-based deep learning models to count objects and generate density maps in complex scenes. Interactive tools enable you to select regions of interest and create exemplar patches as model input.

These capabilities support industrial applications such as defect detection in manufacturing and component counting in automated inspection systems.

Functions

expand all

Create Synthetic Training Data

`insertObjectInImage`	Insert object from source image into image
`objectInsertionDatastore`	Create synthetic labeled datastore for training an instance segmentation or object detection network

Detect Objects

`yoloxObjectDetector`	Detect objects using YOLOX object detector
`detect`	Detect objects using YOLOX object detector
`trainYOLOXObjectDetector`	Train YOLOX object detector

Count Objects

`counTRObjectCounter`	Count objects using CounTR deep learning model
`densityMap`	Density map of object count generated using CounTR deep learning model
`countObjects`	Count objects using CounTR deep learning model
`uiselectboxes`	Select bounding box ROIs from image
`extractpatches`	Extract patches from image at specified ROIs

Topics

Detect Objects Using YOLOX Deep Learning Network
Detect objects for visual inspection and computer vision tasks using YOLOX object detector.
Evaluate Object Detector Performance
Evaluate object detector performance using metrics such as average precision, precision recall, and confusion matrix.
Select Detection Approach for Visual Inspection
Choose among anomaly detection, shape matching, object detection, and object counting for your visual inspection application.

Featured Examples

New

Train YOLOX Object Detector Using Synthetic Data

Train YOLOX object detector using synthetic image data and fine-tune on real data.

Open Live Script

New

Generate Synthetic Training Images Using Copy-Paste Augmentation

Generate synthetic training image data store for training defect detection deep learning models.

Open Live Script

Detect Defects on Printed Circuit Boards Using YOLOX Network

Detect, localize, and classify defects in printed circuit boards (PCBs) using a you only look once version X (YOLOX) deep learning network.

Open Live Script

Detect Small Objects Using Tiled Training of YOLOX Network

Detect small objects in full-resolution images using tiled training of a you only look once version X (YOLOX) deep learning network.

Open Live Script

Count Objects Using CounTR Model

Demonstrates few-shot object counting using the Counting Transformer (CounTR) model, which counts objects in an image by identifying a few exemplars without additional training.