Created attachment 158672 [details] Too many wrong faces Hello, I have the problem that far too many wrong faces are recognized and suggested for the person. I have rejected all suggestions several times and recreated the training data via maintenance. Nevertheless, more than 2000 faces are suggested. Many faces could be filtered out if the person in question has already been assigned in the photo. So if the name "Maik Qualmann" is confirmed in the photo, then it cannot be another person in the photo. It might also help if the automatically assigned faces are grouped by similarity.
We discussed a long time ago whether a person can appear multiple times in the image, through EBV or a mirror image. What accuracy value did you set? Maik
*** This bug has been marked as a duplicate of bug 436544 ***
See attachmant = 80 I've also seen a group photo where a name was suggested 10 to 15 times. Andy
Git commit c003f95327d1239378241226140f9098b989e52e by Maik Qualmann, on behalf of Michael Miller. Committed on 02/11/2024 at 22:56. Pushed by mqualmann into branch 'master'. closing tickets Related: bug 431797, bug 415782, bug 472031, bug 464266, bug 423113, bug 444160, bug 436544 FIXED IN: 8.5.0 M +8 -8 NEWS https://invent.kde.org/graphics/digikam/-/commit/c003f95327d1239378241226140f9098b989e52e