Hi. I saw your example with the fly video which has labels on the repeated black lines on the fly’s back. How does this work? I am using DeepLabCut with fish and I want to put labels on the fin rays, where the number of fin rays on say a dorsal fin can vary from fish to fish. Is there some way to control how the algorithm might do some sort of counting of the rays in relation to each other? How do I know what level of granularity such a morphological point might be classified as in the ImageNet data this uses? Thank you.