SUMMARY In response to https://bugs.kde.org/show_bug.cgi?id=459537, a change was submitted in v8.0 so that "Ignored" face regions are no longer written to the XMP data of the file. I do not agree with this change. Ignored face regions are valuable information that should be persisted with the file along with other face regions. This allows sharing this information with other programs. With the current change in place, this data is not available outside of Digikam. I have spent considerable time on tens of thousands of face regions to identify those that are not of interest. I have also written python code to read these regions and perform various tasks. I would request that this either be reversed, or an option be provided in the GUI for the user to select whether the ignored regions should be written to file. STEPS TO REPRODUCE 1. Tag face region as "Ignored" 2. Write metadata to file 3. Review face regions in file XMP data OBSERVED RESULT - "Ignored" face region(s) are not written to the XMP data of the file EXPECTED RESULT - "Ignored" face region(s) are written to the XMP data of the file SOFTWARE/OS VERSIONS Windows: 11 macOS: Linux/KDE Plasma: (available in About System) KDE Plasma Version: KDE Frameworks Version: Qt Version: ADDITIONAL INFORMATION
Ignored face regions cannot be evaluated in other programs. The metadata standard does not known ignored or unconfirmed faces (these are also not written). There is no known program besides digiKam that also works with ignored faces. What should be written as people's names, English "Ignored" or translated into the user's language? If we save it translated, it will be difficult to import the images because we would have to check in all languages whether the person's name is "Ignored". But you woke up sleeping dogs with this bug report, as the saying goes. I would remove this ignored face feature from digiKam a long time ago. Why ignore dozens of faces in a group photo? Simply delete it and ensure that the image is not easily scanned again. Maik
It is unfortunate that there is no defined way to capture this info in the MWG standard. However, I would observe that other tools support the idea of an "ignored" face. If MWG regions are not suitable for the capture of these in XML, it might be appropriate to write these to custom tags in the DigiKam namespace. I must respectfully disagree with the idea that ignored face regions could be handled by deleting the region and avoiding rescanning. This would be feasible only if the face detection and recognition engines were perfect and/or would never be improved. The reality is that neither off these is true. I for one have spent endless hours reviewing face detection/recognition results in tens of thousands of photos. Knowing that a face is not of interest is important and valuable and should be preserved.