I have big collections of pictures showing human faces, one per picture. All collections were anonymized; the pictures are only named face0001, face0002 and so on. Normally every collection should contain a person exactly once, so I have to sort out duplicates, in terms of two or more shots of the same face (not a simple file compare and not a similarity analysis of the picture since the pictures can originate from different locations, show different haircuts and so on).

The pictures should be ideal for facial recognition. So I hoped for the support of tools like digiKam.
digiKam can scan the whole collection for faces and then try to assign these to already known persons.

And here comes the problem:
digiKam seems to need the user to name every single face before additional runs of the search algorithm can try to find the same person again. That does not help me since my normal case is that there is only one shot per face, not more than maybe a dozen duplicates within 10000 pictures.

So I need something similar like digiKam but without the need of manually tagging the already found, distinct faces.

I have searched the Internet for help concerning my task but could not find some free offline tool to help with it. The facial recognition algorithms of potential tools do not have to work absolutely correct. I have no problem with manually looking through a small preselection of potential duplicates. The tool should just automatically create tags / categories for "similar looking faces".
I do not care about its platform.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.