MVTec HALCON 26.05, launching on May 20, 2026, focuses on improving speed across both AI-based and rule-based machine vision, including faster object detection, more robust data augmentation, and new tools for reading codes on curved surfaces.
A clear focus on speed across industrial vision
MVTec Software GmbH is set to release the latest version of its machine vision platform, MVTec HALCON 26.05, continuing its regular update cycle with a strong focus on performance.
This release is centred on a simple but important goal. Make machine vision systems faster without compromising reliability.
Jan Gärtner, Product Manager for HALCON, explains that the aim is to ensure applications across industrial environments not only remain precise and robust, but also operate at significantly higher speed, especially in demanding automation scenarios.
Faster AI object detection without losing accuracy
One of the standout updates in HALCON 26.05 is the new generation of deep learning-based object detection.
According to MVTec, this delivers:
- Up to 5x faster inference
- High detection accuracy
- Reliable performance across small objects and varying object sizes
The system also includes built-in data augmentation, helping models stay stable when conditions change on the shop floor, whether that is lighting variation, rotation, or partial occlusion.
Reading Data Matrix codes on curved surfaces
A practical addition in this release is the new rectification capability for Data Matrix codes.
In many real-world applications, codes are placed on curved or flexible materials, which can distort their geometry and make them harder to read. HALCON 26.05 addresses this by correcting those distortions before decoding.
This is particularly relevant for:
- Cylindrical components
- Curved packaging
- Flexible materials
The feature can also be integrated into existing workflows, making it easier to adopt without reworking entire systems.
Shape matching becomes more stable with less manual work
HALCON 26.05 also introduces automatic contour optimisation for shape matching.
In practice, reflections, shadows, and textures often create unstable contours that reduce reliability and require manual cleanup. The new approach removes those problematic contours automatically using sample images.
The result is:
- Faster matching
- Improved stability
- Higher accuracy
This is especially useful in environments with reflective or textured parts where consistency is difficult to maintain.
Smarter data augmentation built into the workflow
Another update comes in the way data augmentation is handled.
HALCON 26.05 moves from a procedure-based approach to an operator-based system, allowing developers to define augmentation pipelines directly within their deep learning workflows.
This gives more flexibility and helps:
- Improve model robustness
- Support better generalisation
- Reduce the need for large training datasets
HDevelopEVO preview expands development capabilities
Alongside the main release, MVTec is also introducing a new preview version of its development environment, HDevelopEVO.
With this update:
- Scripts can now be integrated into applications via the .NET interface
- Multimodal LLM support has been expanded
- Developers can use visual prompting, incorporating image data directly into prompts for the AI Assistant
This points toward a more integrated way of developing machine vision applications, combining traditional workflows with AI-driven tools.
Why this matters for manufacturers
What stands out in HALCON 26.05 is not a single headline feature, but a consistent focus on improving how systems perform in real production environments.
Speed improvements in object detection directly impact cycle times.
More robust augmentation reduces the need for constant retraining.
Better contour handling cuts down on manual tuning.
Taken together, these changes are about making machine vision systems easier to deploy and more reliable once they are running.
MTN Analysis
This release reflects where machine vision is heading right now.
There is a clear shift toward making AI systems more practical on the shop floor. Faster inference matters when inspection is part of a high-speed process. Stability matters when conditions are not controlled.
The continued focus on both AI and rule-based methods also reflects how most manufacturers operate today. These systems are rarely standalone. They work together.
The addition of visual prompting in HDevelopEVO is another signal. Development environments are starting to blend traditional engineering with AI-assisted workflows, and that trend is only going to accelerate.
Key Takeaways
- MVTec HALCON 26.05 launches May 20, 2026
- Up to 5x faster AI object detection inference
- New Data Matrix rectification for curved and deformed surfaces
- Automatic contour optimisation improves matching stability
- Enhanced data augmentation integrated into workflows
- HDevelopEVO preview introduces visual prompting and .NET integration
FAQ: MVTec HALCON 26.05
What is MVTec HALCON 26.05?
The latest version of MVTec’s machine vision software, focused on improving speed and performance across AI and rule-based methods.
When will it be released?
May 20, 2026.
What is the biggest improvement?
Faster performance, including up to 5x faster inference in object detection, alongside broader workflow enhancements.
Where is HALCON used?
Across industries such as electronics, semiconductors, battery production, food, agriculture, and logistics.
Find more information here: https://www.mvtec.com
Read More about MVTec
