MVTec Deep Learning Tool 23.11 released

The long-awaited new version 23.11 of the MVTec Deep Learning Tool has been released.

The most important point is that semantic segmentation can now be used not only for annotation but also for learning and evaluation. Semantic segmentation is an approach to learn "defects" and realize "NG" if there are defects. For example, if "flaws" are learned, a common AI model can be used even if the variety changes. A common AI model may also be used for multi-sided inspections.
Until now, we have been using our FIS-200DL to learn a dataset annotated with the Deep Learning Tool and generate a model for evaluation. This time, we have taken the approach of learning good products.
Now, in addition to the "anomaly detection" approach of learning good products, we are able to perform semantic segmentation and evaluation in a consistent manner, and we feel that we finally have all the weapons in our arsenal.

OUR FIS-200DL HAS BEEN CONFIRMED TO WORK WITH EMBEDDED AI MODELS FOR SEMANTIC SEGMENTATION EVALUATED IN DLT23.11.