https://bugs.kde.org/show_bug.cgi?id=444160
Bug ID: 444160
Summary: Facing matching is guess work at best
Product: digikam
Version: 7.3.0
Platform: Microsoft Windows
OS: Microsoft Windows
Status: REPORTED
Severity: normal
Priority: NOR
Component: Faces-Recognition
Assignee: digikam-bugs-n...@kde.org
Reporter: spa...@ecybernard.com
Target Milestone: ---
SUMMARY
I added a collection of 40,000 pictures, and set digikam to the yolo v3 highest
level scanning. It detected a lot of face, after 4 days of scanning.
However, in step 2 of face matching it just didn't match many faces and most of
the ones it did match were not even close.
STEPS TO REPRODUCE
1. Add collection of 40,000 photos
2. set to yolo v3 and multi-core,run face detection wait 4 days
3. Do face matching
OBSERVED RESULT
95% of the results were just plain wrong. I tried adjusting the accuracy
slider, but it just resulted in few results. Using pHash set to 98% usually
results in better matches, but this program doesn't support it.
Google picasa is way faster, and it's matches are mostly right.
EXPECTED RESULT
Facing matching to actually match faces correctly.
SOFTWARE/OS VERSIONS
Windows: 10 build 20H2
--
You are receiving this mail because:
You are watching all bug changes.
