In MERLIC 5.6, some methods that were previously available on a test basis have been transferred to completely new tools. These tools are: “Count with Deep Learning”, “Segment Image Pixel-Precisely", and “Recognize Color”. With “Count with Deep Learning”, also known as Deep Counting, it is possible to count a large number of objects quickly and robustly with little training effort. There are many possible use cases for this feature, like for example completeness checks. “Segment Image Pixel-Precisely" makes it possible to localize trained defect classes with pixel precision.