[digikam] [Bug 338176] New: Digikam-4.2.0 sometimes crashes when recognising faces

[Stephan Olbrich]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #57 from Stephan Olbrich <[hidden email]> ---
Created attachment 91667
  --> https://bugs.kde.org/attachment.cgi?id=91667&action=edit
valgrind output

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Tobias Leupold]

https://bugs.kde.org/show_bug.cgi?id=338176

Tobias Leupold <[hidden email]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[hidden email]

--- Comment #58 from Tobias Leupold <[hidden email]> ---
I ran into the (probably) same problem using KPhotoAlbum: sometimes, training a
face did last for 20 minutes or so with the recognition.db growing by > 100 MB
(I recently wrote emails about this to you and Marcel, but I didn't find the
cause).

With current git master (916b00a), this problem is not reproducible anymore
with my "test" recognition.db I used to provoke this behavior. With libkface
14.12.2, I was able to do so each and every time.

Seems like it's fixed :-)

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Gilles Caulier-4]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #59 from Gilles Caulier <[hidden email]> ---
Tobias,

Great to see this entry solved for you too. This confirm the fix.

Q : The face recognition DB growing is one problem. What's about the memory
consumption on your computer while face recognition ? Did you seen a problem
relevant of memory allocation not free at each commit in DB ?

Note : the memory is not leak. When host application is done, memory is free
properly. It's just a run time problem (if it still exist)

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Tobias Leupold]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #60 from Tobias Leupold <[hidden email]> ---
At least, I did not notice an unusual memory usage. But I also did not search
for it … could also be due to the different ways libkface is used in Digikam
and KPA.

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Gilles Caulier-4]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #61 from Gilles Caulier <[hidden email]> ---
Tobias,

In digiKam Face management is multi-threaded on multi-core CPU.

At least, did you use separated thread in KPhotoAlbum ?

Memory comsuption was visible on my computer when i process large collection.
As i use 16Bb of RAM, i never crash digiKam (memory is never fully used).

Did you see a similar dysfunction on your computer ?

Gilles

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Tobias Leupold]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #62 from Tobias Leupold <[hidden email]> ---
I actually implemented multithreaded training processes, but those were
unusable for our case because the recognition processes were mutexed by the
training. So there was no benefit because the database was locked anyway.

The KPA workflow is apparently quite different from the Digikam workflow: it's
Open image → detect faces → recognize faces → correct falsely detected faces by
training them one by one → open next image → repeat.

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Gilles Caulier-4]

https://bugs.kde.org/show_bug.cgi?id=338176

Gilles Caulier <[hidden email]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED
   Version Fixed In|                            |4.9.0

--- Comment #63 from Gilles Caulier <[hidden email]> ---
To end-users who report this file.

I think bug is fixed with current implementation from git/master (next 4.9.0)

I close this file now. Don't hesitate to re-open if necessary...

Gilles Caulier

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Johannes Kapune]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #64 from Johannes <[hidden email]> ---
new update of digikam 4.9.0 on a openSUSE 13.1

mark a face and name it -> all memory will eaten and crash of digikam but no
crash KDE handler pops up, sorry.

I will try to run it with gdb but last time I was not able to install debug
files

If it is really a Qt problem: what can I do to find out? What version will run
properly? Maybe I have to update some other parts to get it run.
someone can dump a list of used parts (programs) to compare?

Johannes

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Gilles Caulier-4]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #65 from Gilles Caulier <[hidden email]> ---
See last comment from this entry :

https://bugs.kde.org/show_bug.cgi?id=323888

Gilles Caulier

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Johannes Kapune]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #66 from Johannes <[hidden email]> ---
Thanks a lot, yes it works now.
Johannes

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Gilles Caulier-4]

https://bugs.kde.org/show_bug.cgi?id=338176

Gilles Caulier <[hidden email]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[hidden email]

--- Comment #67 from Gilles Caulier <[hidden email]> ---
*** Bug 346097 has been marked as a duplicate of this bug. ***

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[anonykatz]

https://bugs.kde.org/show_bug.cgi?id=338176

[hidden email] changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[hidden email]

--- Comment #68 from [hidden email] ---
Hi there,

got this problem with digikam 4.9, so I updated now to 4.11. But the problem is
still there in 4.11.
This problem seems to be there for a long time now.
Is there are working patch for 4.11?

Greets,
Nadine

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Gilles Caulier-4]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #69 from Gilles Caulier <[hidden email]> ---
Problem is fixed in libkface.

Gilles Caulier

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[anonykatz]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #70 from [hidden email] ---
How can I get the libkface into tarball installation?

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Gilles Caulier-4]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #71 from Gilles Caulier <[hidden email]> ---
libkface is released now with last stable KDE4

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[Gilles Caulier-4]

https://bugs.kde.org/show_bug.cgi?id=338176

Gilles Caulier <[hidden email]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[hidden email]

--- Comment #72 from Gilles Caulier <[hidden email]> ---
*** Bug 350549 has been marked as a duplicate of this bug. ***

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[bugzilla_noreply]

https://bugs.kde.org/show_bug.cgi?id=338176

Gilles Caulier <[hidden email]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[hidden email]

--- Comment #73 from Gilles Caulier <[hidden email]> ---
*** Bug 353859 has been marked as a duplicate of this bug. ***

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[bugzilla_noreply]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #74 from Ritesh Raj Sarraf <[hidden email]> ---
Commenting here since my bug was duplicated to this bug.

Given the dependencies that digikam has, it would be much wiser for digikam to
have an option (or in my opinion, have that as default) to restrict its
consumption of memory to a defined limit.

For CPU and IO, it can be tricky. But please, for memory, you definitely can do
that in-app. There is no good experience to the user, when by the system
reports, it shows that digikam has eaten all your memory. Ideally this should
have been intelligently handled by Linux, but until then, can you have
something in the application itself ?

For most users: if you run into massive memory consumption, please try the
following

prlimit --pid 26286 --rss=2465792000

In the above example the process ID of digikam is restricted memory to 2 GiB.
And my face detection scan worked like a charm after that trick.

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[bugzilla_noreply]

https://bugs.kde.org/show_bug.cgi?id=338176

--- Comment #75 from Ritesh Raj Sarraf <[hidden email]> ---
@Gilles Caulier:

You mentioned fixed in libkface. Can you please specify the version?
I am currently on Digikam  with the bundled libraries and still see the
problem.

digiKam version 4.13.0
CPU cores: 4
Demosaic GPL2 pack support: Unknown
Demosaic GPL3 pack support: Unknown
Exiv2 can write to Jp2: Yes
Exiv2 can write to Jpeg: Yes
Exiv2 can write to Pgf: Yes
Exiv2 can write to Png: Yes
Exiv2 can write to Tiff: Yes
Exiv2 supports XMP metadata: Yes
LibCImg: 130
LibEigen: 3.2.91
LibExiv2: 0.25
LibJPEG: 62
LibJasper: 1.900.1
LibKDE: 4.14.12
LibKExiv2: 2.4.0
LibKGeoMap support: no
LibKdcraw: 2.4.2
LibLCMS: 2060
LibLensFun: 0.2.8-0
LibLqr support: yes
LibPGF: 6.14.12
LibPNG: 1.2.50
LibQt: 4.8.7
LibRaw: 0.16.2
LibTIFF: LIBTIFF, Version 4.0.5 Copyright (c) 1988-1996 Sam Leffler Copyright
(c) 1991-1996 Silicon Graphics, Inc.
Parallelized demosaicing: Unknown
RawSpeed codec support: Unknown
Baloo support: no
Database backend: QSQLITE
Kdepimlibs support: no
LibGphoto2: 2.5.8
LibKface: 3.5.0
LibKipi support: no
LibOpenCV: 2.4.9.1
Sqlite2 support: Yes

--
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Digikam-devel mailing list
[hidden email]
https://mail.kde.org/mailman/listinfo/digikam-devel

[bugzilla_noreply]

https://bugs.kde.org/show_bug.cgi?id=338176

Peter <[hidden email]> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[hidden email]

--- Comment #76 from Peter <[hidden email]> ---
This memory blow-out occurs during face tagging on my digikam 4.0 that came
with Linux Mint 17.3 KDE and also on digikam 4.14 which I installed from
philip5. The dodgy process fills swap memory quickly until it reaches the
maximum crashing the program at that point. Shutting down the digikam GUI
before this happens leaves a digikam process running until the memory limit is
reached. I've read most of the above and the "RESOLVED FIXED" but it's very
confusing. Can someone please give the solution in plain English?

--
You are receiving this mail because:
You are the assignee for the bug.