Karlsruhe, Germany — February 26, 2026 — EVT today announced the release of EyeVision 26.2.1, featuring a new icon set and updated naming designed to make the software more intuitive and consistent across modules. The release also introduces a refined version scheme and a range of stability and usability improvements—helping teams build, maintain, and scale machine vision applications more efficiently.

EyeVision 26.2.1 is built around one goal: make daily work faster and clearer—from navigating the interface to training Deep Learning models and working with 3D overlays.

What’s new in EyeVision 26.2.1

• New icons + updated naming
  • A refreshed icon set for a more modern UI experience
  • Updated naming across parts of the software to improve clarity and consistency
  • Example: users will notice a cleaner visual structure in navigation and dialogs, with clearer labeling that reduces ambiguity during setup and troubleshooting
• New version scheme: YEAR.MONTH.REVISION
  • EyeVision now follows a version format that makes releases easier to track internally and externally
  • Example: 26.2.1 clearly maps to 2026 / February / revision 1

Improvements that users will actually feel

• Deep Learning: better resume training results
  • Object Detection & Image Segmentation training can now save the optimizer state
  • Example: if you pause training and resume later, the model can continue more effectively—often improving the quality of resumed training (with a trade-off of increased model size)
• Process GUI: more flexible, more reliable
  • GUI elements can now show icon + caption simultaneously
  • Example: buttons can display both a symbol and text—improving usability on shopfloor HMIs
• 3D overlays: consistent results across systems
  • Text overlay font sizes are now specified in pixels for consistent rendering
  • Example: identical overlays look the same on different PCs and displays—important for standardized reporting and visual QA
• Connectivity & platform support
  • RTSP streaming support expanded for unix-x86_64 and aarch64 platforms
  • Example: easier deployment on industrial PCs and ARM-based edge devices
• Stability and quality-of-life fixes
  • Improved robustness for image transfer with large overlays and shared memory behavior
  • Result display handling improved (default color options, cleaner frames/background behavior)
  • Reactive UI behavior improved in camera viewer and data tree interactions

EVT also continues to invest in scripting, matching, and performance improvements across recent versions:

• Python Script Interpreter enhancements (Data Tree API access on Unix, performance and stability improvements, geo2d helper classes)
• Smart Match upgrades enabling more robust matching under affine transformations (stretched/skewed patterns)
• 3D tool performance improvements and new parameters that support filtering and controlled tracking direction