This website uses cookies primarily for visitor analytics. Certain pages will ask you to fill in contact details to receive additional information. On these pages you have the option of having the site log your details for future visits. Indicating you want the site to remember your details will place a cookie on your device. To view our full cookie policy, please click here. You can also view it at any time by going to our Contact Us page.

AI technology added to HALCON machine vision software

07 October 2018

MVTec Software GmbH is set to release the new version 18.11 of its standard software HALCON.

Highlights of the new release are said to include the use of AI technologies such as deep learning and convolutional neural networks (CNNs). For example, object-, feature-, or error classes trained with deep learning can now be segmented with pixel accuracy. Similar to classification, segmentation can be performed on a GPU as well as a CPU. 

MVTec provides pretrained networks based on millions of royalty-free images for training the required network. Customers can train new objects more easily as they do not require hundreds of thousands of their own application images. This paves the way for an entirely new range of applications, which previously could not be realised. As the images contained are free of third-party rights, they can be used in commercial applications without reservations.
 
Another feature of the new HALCON version is the localisation of trained object-, feature-, or error classes in an image with the help of deep learning algorithms. In contrast to pixel-precise semantic segmentation, the sought-after objects are each marked in the image by a surrounding rectangle (a so-called bounding box). This leads to potential applications in quality control. A wide range of complex machine vision localisation tasks can be re-evaluated and performed with less effort than conventional methods. 

The object detection inference also runs on both a GPU and a CPU, which offers significant advantages for use in industrial environments. 
 
The new release now also runs out of the box on the 64-bit Arms architecture.
 
A USB3 Vision interface for both 32 and 64 bit is also included. A further improvement is that the reading of data codes has been optimised and made more flexible. For example, the ECC200 data code reader is much faster and capable of reading codes with missing or damaged/disrupted ‘quiet zones’. In addition, codes against complex backgrounds can be found and read faster and more robustly.


Contact Details and Archive...

Print this page | E-mail this page