Existing Core AFR (automatic feature recognition) API takes body as input and always runs feature recognition on the whole body. Local feature recognition (LFR) allows the user to run feature recognition on a limited region of the body. It takes faces from a single body as input in which user is interested and runs feature recognition on that part only. The result is fast and desirable output.
LFR expects the user to provide all the faces of feature that the user wants. But in many cases where single feature consists of several faces, it might be cumbersome for the user to select all faces and run feature recognition. In such cases smart feature recognition (SFR) can be used. SFR doesn’t expect user to input all faces of single feature. It can take any one or more faces of a feature/features and try to recognize features associated with that face smartly.
All faces of a feature for LFR