🐝 Frame side queen cup detection

Problem

We have a very simple queen cup object detection running that was training on small ~15 images in a dataset, its in image-splitter service calling clarifai.com that hosts the model

The problem is that in beekeeping, there are different types of queen cups. We need to distinguish them

Suggested steps

Create and train either two different models, or, better - one model with two classes.