MVTec Software GmbH (www.mvtec.com), a leading international provider of machine vision software, will launch the new version (21.11) of its HALCON machine vision software on November 17, 2021. This version contains many new and optimized features that can be used to implement machine vision applications even more robustly and professionally in a lot of all industrial sectors. Among other things, the new developments include the addition of instance segmentation to the available deep learning technologies, an improved barcode reader, as well as greater usability for dictionaries and Generic Shape Matching. HALCON 21.11 also comes with a plug-in for the OpenVINO toolkit from Intel. By the time of the release, it will also be possible to use the plug-in for other software products from MVTec.
"With HALCON 21.11, we remain in step with the times. The addition of the comprehensive toolbox sets new machine vision standards for a lot of all industrial sectors. We also keep our promise of delivering crucial added value for users through continuous further developments, advanced features, and short release cycles," explains Mario Bohnacker, Technical Product Manager for HALCON at MVTec Software GmbH.
Combining the benefits of semantic segmentation with those of object detection
One highlight of HALCON 21.11 is the addition of instance segmentation technology to the range of deep learning functions. This technology combines the benefits of semantic segmentation with those of object detection and enables the pixel-precise assignment of objects to different classes. It is particularly useful in applications where objects are very close together, touch each other, or overlap. This is the case, for example, when gripping randomly arranged objects from bins ("random bin picking") and when identifying and measuring naturally grown structures, such as organic material.
MVTec has also improved the barcode reader for the 128/GS1-128 code. Thus, it is now also possible to read codes that are blurred due to movement or when the depth of focus is limited. Code 128/GS1-128 enjoys widespread use and is often employed in logistics, due to its compact size and great data density.
Improved usability, faster application development, and more efficient image processing
Another improved feature has to do with the handling of dictionaries, which can be managed even faster and more easily with HALCON 21.11. The dictionaries can now be used with far fewer operator calls, thus speeding up and simplifying the development process. The same holds for improvements to Generic Shape Matching. Based on customer feedback, the usability was increased, for example by automatically determining many more parameters. This makes the access to MVTec's industry-tested shape matching technologies more user-friendly.
With the HALCON 21.11 release, users of the previous version also benefit from the advantages of Intel's OpenVINO toolkit. The corresponding plug-in which can also be used for other MVTec software products in the future, makes it possible to access AI accelerator hardware that is compatible with the OpenVINO toolkit from Intel. This allows deep learning inference to run much faster on Intel processors, including CPUs, GPUs, and VPUs.
Plugin for Intel distribution of OpenVINO toolkit
MVTec software products will now benefit from AI accelerator hardware that is compatible with the OpenVINO toolkit from Intel. As a result, significantly faster deep learning inference times can be achieved on Intel processors including CPUs, GPUs and VPUs for key workloads. By expanding the range of supported hardware, users can now harness the performance of a wide range of Intel devices to accelerate their deep learning applications and are no longer limited to a few specific devices. At the same time, the integration works seamlessly and is not bound to certain hardware specifics. Simply by changing parameters, the inference of an existing deep learning application can now be executed on devices supported by the OpenVINO toolkit.
Generic support for AI accelerator hardware
The OpenVINO toolkit plugin from MVTec is based on the new HALCON AI Accelerator Interface (AI²). This generic interface allows customers to use supported AI accelerator hardware for the inference part of their deep learning applications – quickly and conveniently. Such special devices are widely used especially for applications in the embedded environment, but also exist more and more in the PC environment. By abstracting the deep learning models from specific hardware, the AI Accelerator Interface is particularly future-proof. In addition to plugins provided by MVTec, the integration of customer-specific AI accelerator hardware is also possible. Moreover, it is not only typical deep learning applications that can be accelerated via AI². All "classic" machine vision methods with integrated deep learning functions, such as HALCON's Deep OCR, benefit from this as well.
"Thanks to the seamless integration of the OpenVINO toolkit as an AI² plugin, simply adding supported Intel devices can significantly speed up inference in applications without any further modifications," confirms Thomas Hopfner, Product Manager Licensing & Interfaces at MVTec Software GmbH. "In addition, the use of special accelerator hardware reduces temporal variance and thus leads to deterministic execution times, which is a critical factor especially in the field of manufacturing automation," he adds.
Alex Myakov, Chief Computer Vision Advocate at Intel Corporation adds: "With the release of this plugin, MVTec underlines the increasing market relevance of the OpenVINO toolkit. Through our cooperation, MVTec customers from the industrial machine vision market can now very easily access our wide hardware range of CPUs, GPUs and VPUs and are thus able to implement their deep learning applications even more efficiently."
Vision China
MVTec Software will participate in Vision China Shenzhen this year for the first time with an own booth. From October 28 to 30, 2021, the company will be presenting its latest product releases and innovative machine vision technologies at Booth 9H37. Highlights include the latest features of its standard software versions HALCON 21.05 and 21.11, MERLIC 5, Deep Learning Tool, and state-of-the-art embedded vision solutions. With its presence at the machine vision trade fair in China, MVTec is highlighting the importance of the Chinese market once again. As early as April of last year, the company opened an office near Shanghai in order to be able to better serve this major growth market.
“With our office in Kunshan near Shanghai and our long-standing distributor DAHENG IMAGING, we offer our customers in China optimal support and maximum added value since many years. And with our first-time participation in Vision China Shenzhen, one of the most important trade shows for machine vision, we also give visitors in South China an opportunity to speak directly with MVTec experts,” explains Dr. Olaf Munkelt, Managing Director of MVTec Software GmbH.
Live demos aimed at specific market requirements in China
At the MVTec booth, various technologies and areas of application will be presented that are especially relevant for the Chinese market. Based on a number of live demonstrations, visitors will gain transparent, practical insight into the many different machine vision applications from MVTec. One of the exhibits is a new offline embedded board that integrates hardware and typical cameras from different manufacturers. The board will serve to demonstrate both the various potential applications and the high compatibility of MVTec products with different embedded hardware.
A demonstration of how the software identifies various screws on a rotary table will cover the ways in which HALCON can be used to implement deep-learning-based object recognition. On the basis of practical examples and videos, an interactive touchscreen demo will present a variety of topics and technologies relating to machine vision, including anomaly detection, Deep OCR, the Subpixel Bar Code Reader, and an agricultural application relating to 3D plant inspection. Features of the new version MERLIC 5 all-in-one software will also be presented. In addition, a demonstration based on the example of circuit board inspection will show how MERLIC makes it much easier to apply complex deep learning algorithms by utilizing a camera from our Chinese distributor DAHENG IMAGING. The demo runs on an NVIDIA board and showcases the robustness of the MVTec software on embedded devices.
