- Easy entry into AI – intuitive interface, no programming skills required
- Complete workflow – labeling, training, evaluation, deployment in one tool
- Seamless integration – works smoothly with HALCON and MERLIC
- Flexible editions – free Basic, Professional, and customizable OEM edition
- Future-proof automation – supports PC- and embedded-based deployments
- Full data control – sensitive image data always stays in-house
Automation in modern production environments is becoming increasingly complex. Flexible quality inspection, variable products, and new defect types demand intelligent solutions that go beyond traditional machine vision. Deep learning has emerged as the key technology to address these challenges, enabling applications that were previously not possible and significantly improving the performance of existing ones.
A crucial first step in any deep learning workflow is labeling training data. The quality of labeled data directly determines the accuracy, robustness, and performance of a model. The MVTec Deep Learning Tool was developed to make this step as intuitive and efficient as possible. Thanks to its user-friendly interface, even users without programming knowledge can quickly label images.
Once data is prepared, the tool also supports training and evaluation of networks. The trained models can be seamlessly integrated into MVTec HALCON and MERLIC software products to perform a wide range of deep-learning-based tasks, including object detection, classification, semantic and instance segmentation, anomaly detection, and Deep OCR. This tight integration makes the tool the core of any deep-learning-powered machine vision application.
This year MVTec released Deep Learning Tool 25.04, further expanding the software’s capabilities to support customers throughout the entire image data preparation workflow.
Highlights include:
- Support for HAILO calibration: Enables fast and efficient inference on embedded devices. Models can be updated and deployed with ease, unlocking new workflows for compact, decentralized automation systems.
- Configurable Light Mode: In addition to the existing dark mode, users can now personalize their working environment with a light mode for increased comfort.
- Improved image search: New filter options make it easier to manage and organize large datasets.
These updates streamline productivity and usability, ensuring that the Deep Learning Tool continues to evolve alongside industry needs.
With this version, MVTec also introduced a new edition model to make deep learning accessible for a wide range of users:
- Basic Edition (free): Ideal for beginners and anyone exploring labeling, training, and evaluation for the first time.
- Professional Edition: Designed for companies that want to integrate the tool into their own solutions and redistribute it. It offers a full range of features and seamless integration with HALCON and MERLIC.
- OEM Edition: Tailored for deep integration into customer-specific products. It provides advanced customization options such as white-label branding, custom color schemes, logos, and predefined parameters for training and evaluation.
This flexibility ensures that the right solution is available for everyone—from newcomers to experienced developers.
Deep learning expands the range of what is possible in machine vision for automation:
- Quality inspection: Detecting subtle defects or anomalies that rule-based methods may miss.
- Logistics: Reading variable characters and codes using Deep OCR.
- Electronics manufacturing: Supporting complex inspection and segmentation tasks.
- General manufacturing: Enabling flexible classification of variable products.
By combining deep learning with traditional machine vision methods, users can create robust hybrid solutions that deliver both transparency and flexibility.
Conclusion
With its newest version, the MVTec Deep Learning Tool lowers the barriers to entry for deep learning in automation while providing advanced features for professionals. The new editions — Basic, Professional, and OEM — ensure that there is a fitting solution for every use case.
For automation specialists, this means:
- Faster development cycles
- Intuitive workflows
- Robust and adaptable models
- Seamless integration into existing systems
- Full data sovereignty
As machine vision continues to evolve, the MVTec Deep Learning Tool provides a powerful foundation for building the next generation of automation solutions.
