Face Recognition in DK 8.5 - are ignored faces learned?

[Gilles Caulier]

Hi Michael,

Do we need to add something in the online documentation about this topic ?

Best
Gilles

Le dim. 12 janv. 2025 à 22:48, Michael Miller <michael_miller at msn.com> a
écrit :

> Hi Steve,
> Your suspicions are correct. Ignoring a face does nothing to face
> training. We don't save the ignored face data for any type of matching or
> training.
>
> We thought of, and I experimented with matching ignored faces to make it
> easier to ignore similar faces. Ultimately, we rejected it because a false
> positive match to an ignored face could cause you to miss an otherwise
> important match to a real person.
>
> Cheers,
> Mike
>
>
>
> Sent from my T-Mobile 5G Device
>
>
>
> -------- Original message --------
> From: Steven Robbins <steve at sumost.ca>
> Date: 1/12/25 4:15 PM (GMT-05:00)
> To: digikam-users at kde.org
> Subject: Face Recognition in DK 8.5 - are ignored faces learned?
>
> Hi,
>
> I'm using the new default recognition settings: Accuracy = 7, Model =
> SFace.
> And I'm very pleased with the results.
>
> I have a question about the "Ignored" tag.  I've chosen to tag only family
> and
> close friends.  So when there is a gathering of another kind, I may find
> 10
> other people in the background of 20+ images.
>
> My question is: if I tag one image to ignore all those people, will a
> subsequent run of face recognition find a match, and see that it is
> ignored,
> and then go ahead and mark that face as ignored?
>
> My experience is that the answer is "no, that's not how it works".  I
> generally have to ignore all those background people in all the individual
> images.  If that's accurate, then: is the described alternate behaviour
> feasible and useful?
>
> Thanks,
> -Steve
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/digikam-users/attachments/20250113/e36464ca/attachment.htm>