MVTec’s HALCON 26.05, is built around one priority: faster processing. Whether it’s AI-based detection or classical machine vision, the update targets throughput, a key constraint as automation lines push for higher speeds and tighter tolerances.
The most important nuance is a fivefold gain in AI-based object detection speed, without touching accuracy. The rebuilt system also handles small and variable-sized objects more reliably, and built-in data augmentation improves resilience to changes in lighting, rotation, and partial obstructions.
Beyond speed, two targeted capabilities stand out. A new Data Matrix code rectification function compensates for geometric distortion before decoding, solving a persistent problem when reading codes on curved or flexible surfaces. And for shape matching, automated contour optimization filters out unstable edges caused by reflections or surface texture, making matches faster and cleaner on difficult materials.
MVTec is also previewing HDevelopEVO, its updated development environment. Developers can now integrate scripts via .NET and use visual prompting with multimodal language models — feeding images directly into AI assistant prompts to speed up application development.
