Hi, question the learning and clustering algorithm used by LR for face recognition.
If I tag photos of people taken at different ages with the same name, does this skew LR's ability to accurately detect similar faces?
E.g. photos of my kids as baby's and toddlers and teenagers, or my friends as teenagers and now as adults.
Will I get better accuracy by applying names based on the faces looking alike, e.g. John Doe Baby, John Do Teenager?
Or will LR remain accurate and cluster similar looking faces even if the clusters end up looking completely different?
The information I was given some time back is that accuracy won't be adversely effected by tagging the same person at different ages. In other words, Lightroom allows tagging of different looking faces of the same person, as he/she grows up over time.
It'd be nice if future versions took the photo metadata into account so once LR gains confirmation that a person was an infant in 2005, it won't continue to suggest the person as a senior citizen in other 2005 photos or even in 2015 photos. This will never work if the algorithm doesn't perceive age-related facial features, though.