Not my term but . . .basically it uses macro detection or classification of various levels vs the micro detection and classification of deep learning. Lots of UAS and mining or slow speed ground vehicle shuttles do this. That ranges from no object classification to macro classification like human, animal etc. My personal preference is to determine what levels are needed based on use case. One where deep learning is used as little as possible due to its untenable inefficiencies of learning, the processing required in real-time and getting confused. In mining for example Imay choose to stop for an object and not care if it is a human. Or I may want to macro classify and run macro prediction. And finally there are cases I need to learn exactly. Like rescue vehicles, signs and uniforms or items of clothing a person wears. like a the police or a crossing guard.