[Digikam-users] Face tagging?

[Gilles Caulier]

2012/3/20 John Stumbles <john at stumbles.org.uk>:
> On 20/03/12 18:37, Gilles Caulier wrote:
>>
>> Note that about Face Management :
>>
>> 1/ Face detection work fine (use OpenCV library and libface)
>> 2/ Face Recognition is not fully implemented in libface. A GsOC 2012
>> project have been open to find a student to work on it this summer :
>>
>> http://community.kde.org/GSoC/2012/Ideas#Project:_Face_Recognition
>
>
> That's great news!
>
> (Doesn't Google's Picassa have face recognition - or is it just detection?)
>
>
>> 2012/3/20 Kim<kim.mills at wightman.ca>:
>>>
>>> I finally figured out how to set up 2.5 and it worked much better. :)  It
>>> took just a few min to scan the folders and it looks like it found all
>>> the
>>> people ones.  I don't mind having to tag them by hand.  It's just a big
>>> help
>>> having them already sorted!
>
>
> I have thousands so doing them by hand is not realistic! And DigiKam's
> interface seems to run extremely slowly in face mode, e.g. just when
> dropping non-faces that it's identified.

In 2.6, we have improved multi CPU support. If you have multicore cpu
this will be better.

Don't forget that main algorithm use OpenCV lib. Optimization are
important here, especially if library is compiled properly, but
digiKam team cannot manage it. So optimization of openCV is delegate
to packagers...

Gilles Caulier