suggestion: using facenet/openface

[Tobias Hermann]

Hi,

having a look at the digiKam source code on github, I see that eigen/fisher
faces and the lbph model from OpenCV are used as face recognition
techniques.

However the deep convolutional neural network architecture "facenet" (
arxiv.org/abs/1503.03832) seems to be the state of the art. With "openface"
(github.com/cmusatyalab/openface) there is a good implementation available.

Have you thought about trying to use it to improve the face recognition
capabilities of digiKam?

Regards,
Tobias
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/digikam-devel/attachments/20170810/c408933f/attachment.html>