MVTec Deep Learning Tool 25.12: Execute Deep Learning applications with ease 

MVTec Deep Learning Tool 25.12

MVTec has boosted its AI machine vision Deep Learning Tool with the release of the Deep Learning Tool 25.12. The new version focusses on improved data handling, increased flexibility during training, and a wide range of usability and stability improvements. 

The Deep Learning Tool Integrates seamlessly with HALCON and MERLIC, supporting a wide range of tasks including object detection, segmentation, anomaly detection, and Deep OCR.

The Deep Learning Tool is available in Free, Professional, and OEM editions. Version 25.12 introduces improvements across all editions, with additional training and model management capabilities available in the Professional and OEM editions. 

Deep Learning Tool Free Edition:

The free edition of the Deep Learning Tool includes everything required to get started with your Deep Learning application: Data management, labelling, training and evaluation. With version 25.12, the Free version adds improved image management through the new tagging system, the ability to move and rename images directly within the tool, enhanced navigation and gallery handling, and general user interface improvements. Stability fixes and usability enhancements included in this release also apply to the Free version.

Deep Learning Tool Professional Edition

The Deep Learning Tool Professional edition allows you to make the tool part of your solution, with a permanent license and the right to resell it. The Professional version is intended for productive use in development and engineering environments. In addition to all improvements available in the Free version, version 25.12 adds support for loading and continuing training of pretrained models from previous HALCON versions. The Professional version also benefits from extended training configuration options, including adjustable or disabled validation to speed up training, and the ability to load external deep learning models for training.

Deep Learning Tool OEM Edition

The OEM edition of the Deep Learning Tool enables deeper integration through white labelling options such as the configuration of colour themes, custom logos and the presetting of training and evaluation parameters. The OEM version is designed for system integrators and industrial deployment scenarios. Version 25.12 includes all improvements available in the Free and Professional versions and further benefits from the advanced training, evaluation, and model handling capabilities required for integration into production systems. Stability improvements in this release particularly address long running workflows, device handling, and export reliability in OEM environments.

Labelling, Training, Evaluation

Highlights of what’s new 

Smarter image management with tags 
Tags can now be created and assigned to images. A tag is a user defined text string that can be used as a filter criterion. This provides a flexible way to organize datasets and simplifies handling large image collections across different applications. 

Reuse and retrain existing models 
Users of the Professional and OEM versions can now load pretrained models from previous HALCON versions. This allows existing models to be maintained and retrained even if they have been replaced in newer versions of the Deep Learning Tool. 

Move and rename images directly in the tool
Images can now be moved and renamed directly from the Gallery and Image pages. These changes are applied to the files on disk or on a file share, removing the need to switch to another application for file management.

Faster training with adjustable validation 
In the Professional and OEM versions, the validation rate during training can now be adjusted or fully disabled using the advanced training parameters. This makes it possible to speed up training, especially during early experimentation phases. 

Additional enhancements 
The light theme has been improved and fonts are now consistent between the user interface and documentation. Evaluation results provide clearer information when label classes in a project differ from those used in trained models. Navigation has been improved with keyboard shortcuts such as Ctrl+G to jump to a specific image index. When switching between images, the current zoomed area is preserved. Dataset splits can now be created based only on the active filter result.

Version 25.12 includes many bug fixes and improvements across classification, segmentation, anomaly detection, and Deep OCR projects. These include fixes for training and evaluation inconsistencies, improved handling of devices and configurations, more reliable exports, and clearer documentation. Stability has been improved, particularly for OEM and production focused workflows. 

Summary 
MVTec Deep Learning Tool 25.12 is a maintenance focused release that improves efficiency, reliability, and clarity when working with deep learning based machine vision projects. The release delivers meaningful improvements for users of all editions while extending advanced training and model management capabilities for Professional and OEM use cases. 

Need a price or more application information? Please email Adept Turnkey or call our offices
Adept Turnkey Pty Ltd are "The Machine Vision and Imaging Specialists" and distributor of MVTec products in Australia and New Zealand. To find out more about MVTec, please call Adept Turnkey at Perth (08) 9242 5411 / Sydney (02) 9905 5551 / Melbourne (03) 9384 1775 or contact us online.