Maybe someone else is in this situation too. But while I'm in the process of training my own version Yv8 I did discover that some of the built in class designations in Yv8 we're so consistently wrong that I could them to good use in the following way. I built a Yv8 'PickerNamer' .py script that allows me to do two things. First, I can select which of the 80 classes I want to display, and second, I can edit the name that's displayed in my object detection engine. For example, Yv8 does not have a class for fish. Yet, fish consistently class-out as birds or kites. So using my PickerNamer, I select bird and kite change their names to fish and voilà. I have a pretty darn good object detector for fish using Yv8. if this description isn't enough for you to build your own PickerNamer, I am happy to share the code for PickerNamer to anyone who reaches out.