TL;DR:
- MVTec Software GmbH will launch version 23.05 of HALCON, its machine vision software, on May 23, 2023.
- The new release focuses on deep learning methods, including Deep Counting, which enables robust counting of large quantities of objects.
- HALCON 23.05 introduces improvements for Deep OCR and 3D Gripping Point Detection.
- Users can optimize deep learning networks for their own applications, enhancing recognition rates and gripping surface detection.
- The update also includes helpful improvements, such as easier integration of external code into HALCON.
- Jan Gärtner, Product Manager HALCON at MVTec, highlights the increased interest in deep learning methods among customers.
- Deep Counting allows quick and accurate counting of objects, even for highly reflective and amorphous materials.
- Deep OCR now offers fine-tuning text detection, resulting in more robust results and improved readability in noisy environments.
- 3D Gripping Point Detection allows robust detection of gripping surfaces, and users can retrain the model with application-specific data.
- HALCON 23.05 introduces the Easy Extensions Interface for seamless integration of external programming languages, expanding functionality.
Main AI News:
MVTec Software GmbH, a renowned global software manufacturer specializing in machine vision, is set to unveil version 23.05 of its flagship machine vision software, HALCON, on May 23, 2023. This release places a strong emphasis on deep learning techniques, with the spotlight on a groundbreaking feature known as Deep Counting. Leveraging the power of deep learning, this method enables the accurate and efficient counting of vast quantities of objects. Moreover, HALCON’s latest iteration incorporates significant enhancements for two key deep learning technologies: 3D Gripping Point Detection and Deep OCR.
HALCON 23.05 introduces the ability to fine-tune the underlying deep learning networks, which come pretrained on industry-specific images, for customized applications. This empowers users to achieve even higher recognition rates for Deep OCR applications, ensuring exceptional accuracy when deciphering texts, regardless of their orientation or font. Additionally, it enables the detection of text with diverse printing styles and previously unseen character types, even in challenging environments characterized by noise and low contrast.
For applications involving 3D Gripping Point Detection, HALCON now offers the capability to retrain the pretrained model using application-specific image data. This advanced functionality ensures the more robust identification of gripping surfaces on any suction-compatible object. The process of labeling the grippable surfaces is made effortless and efficient through the intuitive MVTec Deep Learning Tool.
HALCON 23.05 also introduces the Easy Extensions Interface, a remarkable feature that facilitates seamless integration of external programming languages. This allows customers to extend HALCON’s capabilities beyond traditional image processing and embrace a wider spectrum of functionalities. With the simplified integration of external code, leveraging the power of the .NET framework has become significantly more accessible.
Users can effortlessly incorporate their own .NET code into HDevelop and HDevEngine, benefiting from the extensive array of functionalities offered by the framework. Additionally, they can harness the familiar data types and HALCON operators from the HALCON/.NET language interface, elevating the flexibility and expanding the application possibilities of HALCON.
Jan Gärtner, the Product Manager of HALCON at MVTec, highlights the surge in customer interest surrounding the integration of deep learning methods into their solutions. To address this demand, MVTec meticulously developed the latest HALCON version, introducing groundbreaking deep learning technologies and refining existing features to deliver unparalleled precision and results for customers.
Conlcusion:
The launch of HALCON version 23.05 by MVTec Software GmbH, with its focus on deep learning methods and significant improvements in Deep Counting, Deep OCR, and 3D Gripping Point Detection, signals a significant advancement in the machine vision software market. This release empowers businesses to leverage cutting-edge technology for more accurate and efficient object counting, text recognition, and gripping surface detection.
The ability to optimize deep learning networks for specific applications enhances recognition rates and opens up new possibilities for customization. Furthermore, the Easy Extensions Interface enables seamless integration of external code, expanding the functionalities of HALCON. Overall, these developments reinforce MVTec’s commitment to meeting the evolving needs of the market and providing businesses with advanced tools to drive innovation and success in the field of machine vision.
