The highlights of the new MERLIC 5 release include modern deep learning features and improved camera handling. Along with the latest release, MVTec also offers a new licensing model for maximum flexibility. It allows customers to choose whatever best meets their requirements from a range of different packages and price categories. "With MERLIC 5, we offer high-quality yet easy-to-use machine vision software for all conceivable applications and industrial sectors. It allows our customers to conveniently integrate a machine vision package into their process environment and access innovative deep learning technologies," explains Christoph Wagner, Technical Product Manager for MERLIC at MVTec.
Robust anomaly detection with deep learning
MERLIC 5 contains new and optimized features that further increase usability, robustness, and the field of possible applications. Among these features is included the improved camera handling using the Image Source Manager (ISM), which completely separates the image acquisition source from the image processing part. As a result, machine vision applications can be easily copied and run between different systems without adjusting the camera settings. The graphical user interface allows the image sources to be comfortably managed and configured.
In addition, the release of the new version means that MVTec's latest deep learning functions are now also available in the MERLIC 5 all-in-one machine vision software. It’s now easier than ever for users to exploit the benefits of this technology directly in their machine vision applications. Among other things, MERLIC 5 includes deep-learning-based anomaly detection, which provides important added value, particularly in applications where few or no “bad images” are available. Just a small number of "good images" is sufficient for training so that all types of defects can then be located reliably and precisely. The training process usually takes only a few seconds, meaning that users can perform many iterations to fine-tune their applications within a short period of time. The second deep learning feature integrated into MERLIC 5 is the "classify image" tool. This tool can be used to assign images to trained classes based on deep learning algorithms. The technology is particularly suitable for classifying defects or objects with highly variable shapes and appearance, as in the case of natural products. The models can be trained using, among other things, the free Deep Learning Tool from MVTec and then imported into MERLIC 5.The MERLIC 5 launch will be accompanied in its early phase by a campaign to generate curiosity about its deep learning features and also to help users understand them better.
More flexible use
Machine vision applications are now being used in a growing number of industries and sectors, resulting in an expansion of the fields of application and of the accompanying demands on the machine vision software. In concrete terms, this means that a wide range of image acquisition sources and technical features are required in practice. MERLIC 5's new package-based licensing model meets these very requirements. The packages are designed in such a way that users can choose the exact range of functions that they actually need for their application. "With our new licensing model, we are increasing MERLIC's flexibility, which in turn expands its possible field of applications," explains Christoph Wagner.