https://bugs.kde.org/show_bug.cgi?id=524202

            Bug ID: 524202
           Summary: unique face is confirmed in image; recognition run
                    again finds more occurrences of the same unique face
    Classification: Applications
           Product: digikam
      Version First 9.1.0
       Reported In:
          Platform: Microsoft Windows
                OS: Microsoft Windows
            Status: REPORTED
          Severity: normal
          Priority: NOR
         Component: Faces-Recognition
          Assignee: [email protected]
          Reporter: [email protected]
  Target Milestone: ---

DESCRIPTION
People/Face tagging: image with multiple faces; image with at least one
existing/confirmed face of "Jane Doe"; Recognition finds a second/third/...
face of "Jane Doe" in same image

STEPS TO REPRODUCE
1.  tag faces in images with multiple faces
2.  confirm accurate tags
3.  recognition finds more of the same face in same image when run

OBSERVED RESULT
I see faces recognized as a face already present in the image and confirmed. 
While I understand mirror images, images containing 2D pictures(a picture of a
picture), and reflections, may produce these results - accurately, it would
seem beneficial to assume with a recognition setting of 8, that one "Jane Doe"
face per image would disqualify the "Jane Doe" face tag for additional
recognition in THAT image. ...or at least flag it as redundant or something. 
...or, is this by design?

EXPECTED RESULT
a "twin" feature to differentiate first positive hit from subsequent hits OR a
count of the unique face tag suggestion if >1 OR multiple unique face
occurrence presence to alert user OR ??

SOFTWARE/OS VERSIONS
Operating System (no Kinfo on Windows):  ver=10.0.19045.6456 
KDE Plasma Version: unknown
KDE Frameworks Version: 6.26.0
Qt Version: 6.11.0

ADDITIONAL INFORMATION
I read 472416.  But, it might be easy to just flag occurrences of images with
multiple single unique face hits with a badge on the thumbnail or a number of
repeated faces or something the user could react to as opposed to (what appears
to be) the sought after objective of automating differentiation of a live face
vs a picture in the image/reflection.  if you read it and think it's addressed
I'll pipe down.  It's a great package so far!  thx

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to