gtk-facemanager refuses to capture

[abdulhannankhan]

gtk-facemanager refuses to capture

[GoogleCodeExporter]

could you check /home/$username/.pam-face-authentication/train/
see if there any images in this directory
if files exist then
I came across the same error
its because the loader for pgm format which comes with gtk,  
libpixbufloader-pnm.so ,
is not "recognized" or does not exist
we have to explore this , please report more details about the error(if this is 
not
the case) , update the gtk package also

Original comment by [email protected] on 14 Apr 2009 at 8:12

[GoogleCodeExporter]

Yes, there are a handful of .pgm files - when I tried to view the directory 
gthumb,
it couldn't seem to handle the files.  ImageMagick's display, along with gimp 
show
all the images as small black rectangles.  ImageMagick's identify gave the 
following
(for all the files):
0414090951151.pgm PNM 120x140 120x140+0+0 8-bit Grayscale DirectClass 65.8kb

slocate libpixbufloader-pnm.so:
/usr/lib32/gtk-2.0/2.10.0/loaders/libpixbufloader-pnm.so
/usr/lib64/gtk-2.0/2.10.0/loaders/libpixbufloader-pnm.so

Using GTK gtk+-2.14.7 (everything in my system is up to date to the newest 
stable
version in Gentoo's repository with the occasional testing version)

Original comment by [email protected] on 14 Apr 2009 at 8:30

[GoogleCodeExporter]

can you modify pgm to jpg in the source files and recompile ?

do a grep -R 'pgm'* in the source dir
and replace it with jpg

this should work

Original comment by [email protected] on 14 Apr 2009 at 8:46

[GoogleCodeExporter]

Replacing pgm with jpeg works, but now I still have the problem that all of the
images are black rectangles:
% identify *
04140912175710.jpg JPEG 120x140 120x140+0+0 8-bit PseudoClass 256c 535b

I checked with diff, they are completely identical.

Original comment by [email protected] on 14 Apr 2009 at 9:20

[GoogleCodeExporter]

black images are probably because , your face is not close to the webcam and , 
eyes
are not properly localized
can you take pics looking at the cam very closely

this is a bug which has to be fixed

Original comment by [email protected] on 14 Apr 2009 at 9:38

[GoogleCodeExporter]

yes, moving close fixed the black images.  how well does it successfully
authenticate?  I tried it several times and was only successfully identified 
once. 
does it work better with more training images or less?  any other tips to get 
it to
authenticate successfully?

Original comment by [email protected] on 14 Apr 2009 at 12:52

[GoogleCodeExporter]

yes , you need more examples
4 for each illumination would work
i have 36 pics ,each 4 taken at different time the day
best is too remove all those images which have background(wrong capture)
have all the faces(straight look would be better) and eyes on all images should 
be at
the same level

Original comment by [email protected] on 14 Apr 2009 at 1:35

[GoogleCodeExporter]

can you give your mail id, i will send screenshots of how my face database 
looks like

Original comment by [email protected] on 14 Apr 2009 at 1:52

[GoogleCodeExporter]

It's @gmail.com, thanks

Original comment by [email protected] on 14 Apr 2009 at 3:00

[GoogleCodeExporter]

Use Latest Version of pam-face-authentication

Original comment by [email protected] on 8 Jul 2010 at 10:53

• Changed state: Fixed