For automated inspection and production

Object recognition with machine vision software

Object recognition enables machine vision systems to identify, classify, and locate objects in images or 3D data. It is fundamental to many automated inspection and production tasks – from distinguishing product variants and counting components to completeness checks and robotic handling. Depending on the application, both classical machine vision and deep-learning-based methods can be used.

What do you need to recognize?

From detection to completeness checks

Core tasks of object recognition

Object recognition includes a range of machine vision tasks in both 2D and 3D. 2D object recognition uses visual characteristics such as shape, texture, contrast, or appearance, whereas 3D object recognition also considers spatial information to recognize objects and determine their position and orientation. Depending on the application, objects can be detected and located, classified, counted, or checked for completeness. This makes object recognition suitable for applications such as inspection, sorting, automated handling, and assembly.

Object detection

Detect and locate objects in 2D or 3D, even when their position is unknown or you need to find multiple objects.
Explore deep-learning-based object detection

Classification

Assign objects or images to predefined classes to distinguish products, components, or variants.
Explore deep-learning-based classification

Object counting

Count objects in an image, for example to check completeness in packaging or assembly processes.
Explore deep-learning-based counting

FROM AGRICULTURE TO AUTOMATION

Typical applications

Object recognition is used throughout industrial automation wherever products, components, or other objects need to be identified, distinguished, counted, or checked.

Detecting weeds for selective herbicide application in agriculture.
Checking correct fill levels of biscuit bags in food production.
Commissioning and identifying goods in logistics and automated warehouses.
Sorting and packing pills and medical products in the pharmaceutical industry.
Verifying the correct loading of workpiece carriers in automotive production.
Ensuring proper PCB assembly in electronics manufacturing.
Classifying bacteria cultures in healthcare and science.
Sorting materials such as plastic, wood or metal in recycling applications.

Please note: Once you watch the video, data will be transmitted to Youtube/Google. For more information, see Google Privacy.

ACCURATE & ROBUST

Object recognition technologies in
MVTec software

MVTec software provides classical and deep-learning-based methods for object recognition in 2D and 3D. Depending on the application, these technologies can be used individually or combined.

Classical machine vision

Object recognition & template matching
Rule-based classification
Morphology
Blob analysis

Deep learning

Deep-learning-based classification
Deep-learning-based object detection
Deep-learning-based semantic segmentation

3D object recognition

3D object processing
3D shape matching & 3D surface matching
Camera calibration & hand-eye calibration
Generic Box Finder

More about our technologies

OBJECT RECOGNITION APPLICATIONS

Our success stories

Automated wire harness manufacturing: machine vision gives robots the vision they need

Manufacturing complex wire harnesses is one of the most challenging tasks in industrial automation. Flexible cables constantly change position during processing, making them difficult to handle reproducibly. Together with TE Connectivity, Cellios has…