I would go about it the following way. I’d use the Python bindings for OpenCV, mainly because that is what most of my other outside of OH automation code is written in and put all the recognition in a separate Python service. When a detection event occurs I would publish this to MQTT using the Paho library.
I don’t know much about OpenCV, but since I’ve already written somewhat of a framework for supporting this sort of thing in Python I would probably try to make it a module on that.
Even if I risk the old warning “if all you have is a hammer, all problems look like nails.”