We at "helping faceless" are facing the same issue. Just an introduction to what we do, Every 8 minutes a child goes missing in India. We are trying to help these children with face recognition, Crowdsourced and Government Data.
Essentially We can't extend our search beyond age of 16 and below age of 9, reliably because the face structure changes ever so slightly that our similarity scores explode.We are using 4sf from openbr right now, but hope to look deeply into Facebook's Deep face for face recognition purpose. If only I could find a more detailed paper for deep face.
Essentially We can't extend our search beyond age of 16 and below age of 9, reliably because the face structure changes ever so slightly that our similarity scores explode.We are using 4sf from openbr right now, but hope to look deeply into Facebook's Deep face for face recognition purpose. If only I could find a more detailed paper for deep face.
URL's:
Helping Faceless: http://www.helpingfaceless.com/
4sf: http://www.cse.msu.edu/biometrics/Publications/GeneralBiomet...
Deep Face: https://www.facebook.com/publications/546316888800776/
As a sidenote, everything is open-source and far from perfect. So if any ones to pitch in, help us out, that would be super awesome.