Bug 463350 - Red eyes correction on windows 10 not functionning
Summary: Red eyes correction on windows 10 not functionning
Status: RESOLVED FIXED
Alias: None
Product: digikam
Classification: Applications
Component: Faces-RedEyes (show other bugs)
Version: 7.9.0
Platform: Microsoft Windows Microsoft Windows
: NOR normal
Target Milestone: ---
Assignee: Digikam Developers
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2022-12-22 18:11 UTC by Bob06
Modified: 2023-10-19 06:30 UTC (History)
2 users (show)

See Also:
Latest Commit:
Version Fixed In: 7.10.0
Sentry Crash Report:


Attachments
before and after red eyes correction on windows 10 (30.21 KB, application/pdf)
2022-12-22 18:11 UTC, Bob06
Details
original file (892.72 KB, image/jpeg)
2022-12-22 19:54 UTC, Bob06
Details
corrected file with red-eyes windows 10 (552.45 KB, image/jpeg)
2022-12-22 19:55 UTC, Bob06
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Bob06 2022-12-22 18:11:50 UTC
Created attachment 154759 [details]
before and after red eyes correction on windows 10

OBSERVED RESULT
after a very long time (34 sec) you obtain  an half correction on windows 10.

The correction works fine (normal speed ) on ubuntu 16.04 with the same version 7.9.0

See file attached

SOFTWARE/OS VERSIONS
Windows: 10
Comment 1 caulier.gilles 2022-12-22 18:56:36 UTC
Can you share the image not corrected to try to reproduce the problem here ?

Gilles Caulier
Comment 2 Bob06 2022-12-22 19:54:08 UTC
Created attachment 154760 [details]
original file
Comment 3 Bob06 2022-12-22 19:55:32 UTC
Created attachment 154761 [details]
corrected file with red-eyes  windows 10
Comment 4 Maik Qualmann 2022-12-22 20:22:33 UTC
That's strange, it doesn't work for me either in AppImage-7.9.0 or in the current developer version, the result is like Windows 10. If you set a very low value, you can see that the correction seems to be in the hairs of the young man is applied. The location and distance indicate that the position is not correct.

Maik
Comment 5 Maik Qualmann 2022-12-23 07:19:33 UTC
Git commit d81b9f3fb753cf5830d99fe66fffd3bcc99b4e97 by Maik Qualmann.
Committed on 23/12/2022 at 07:18.
Pushed by mqualmann into branch 'qt5-maintenance'.

use standard face detection model
in the first step for finding faces in Red-Eye Correction.
FIXED-IN: 7.10.0

M  +2    -1    NEWS
M  +18   -10   core/libs/dimg/filters/redeye/redeyecorrectionfilter.cpp

https://invent.kde.org/graphics/digikam/commit/d81b9f3fb753cf5830d99fe66fffd3bcc99b4e97
Comment 6 Bob06 2023-01-03 14:39:27 UTC
(In reply to Maik Qualmann from comment #5)
> Git commit d81b9f3fb753cf5830d99fe66fffd3bcc99b4e97 by Maik Qualmann.
> Committed on 23/12/2022 at 07:18.
> Pushed by mqualmann into branch 'qt5-maintenance'.
> 
> use standard face detection model
> in the first step for finding faces in Red-Eye Correction.
> FIXED-IN: 7.10.0
> 
> M  +2    -1    NEWS
> M  +18   -10   core/libs/dimg/filters/redeye/redeyecorrectionfilter.cpp
> 
> https://invent.kde.org/graphics/digikam/commit/
> d81b9f3fb753cf5830d99fe66fffd3bcc99b4e97

Hello I have downloaded  7.10.0 and the problem is not fixed. 
Only a part of the eyes  is correct.
Comment 7 Maik Qualmann 2023-01-03 20:50:44 UTC
Tested on Windows10 with digiKam-7.10.0-20230103T193317-Win64.exe and the sample image from here. The red eyes are completely corrected. I can't reproduce any problem.

Maik
Comment 8 Bob06 2023-01-04 10:43:56 UTC
(In reply to Maik Qualmann from comment #7)
> Tested on Windows10 with digiKam-7.10.0-20230103T193317-Win64.exe and the
> sample image from here. The red eyes are completely corrected. I can't
> reproduce any problem.
> 
> Maik

 I apologize,  I have not download the good date of compilation.
With digiKam-7.10.0-20230103T193317-Win64.exe it's OK 
Thanks  Bob
Comment 9 Maik Qualmann 2023-10-19 06:30:45 UTC
Git commit d198300bf06951b9a03951af44912525f28a5165 by Maik Qualmann.
Committed on 19/10/2023 at 08:28.
Pushed by mqualmann into branch 'master'.

use YoloV3 face detection again for better red eye detection
Adjust the size of detected faces for problematic
images that didn't work with YoloV3.
Related: bug 384614

M  +12   -10   core/libs/dimg/filters/redeye/redeyecorrectionfilter.cpp

https://invent.kde.org/graphics/digikam/-/commit/d198300bf06951b9a03951af44912525f28a5165