SUMMARY I have a big collection of pictures, with thousands of different people on it. I would actually like to only find a specific group of people in all those pictures (e.g. those I already have in my face tags, or a subset of them). Otherwise I would have to sift out thousands of faces that I have no interest in. SOFTWARE/OS VERSIONS Windows: macOS: Linux/KDE Plasma: (available in About System) KDE Plasma Version: KDE Frameworks Version: Qt Version: ADDITIONAL INFORMATION
To clarify: what I would like is to detect all faces, recognize only those I already have in my face-tag list (or a subset thereof), and automatically not keep all the others in the "unknown" face group.
I do not understand the wish in this bug report. It is possible to filter or search for certain (facial) tags. It is also possible to search only images that contain tag1 and tag2 and tag3 (group of people). Maik
O.k., another try: In all my pictures, I have thousands of different people in the background that I'm not interested in. I don't want those faces to be detected, and I don't want to sift through all those faces to get to the faces of about 20 people I'm interested in. What I would want is this: When initiating a face detection/recognition process, I would like to limit the search to only the faces I am interested in (e.g. family, friends, clients, whatever). Steps: 1) tag the faces you want manually in some of the pictures, as a sample 2) perform the face detection/recognition task, limiting it to only detect and recognize the faces you have set as samples. Am I clearer now?
Ok, but in order to do so Digikam would still need to identify the rest of the faces, and purposely ignore them, since it can't know if a person isn't someone in particular until he/she has been identified. So what you propose is an option to hide or ignore all faces except those that match some specific persons, right? In that case, it would be a step-up from what is proposed in bug #392023, in which you could manually ignore selected faces. But in this case, automatically if they don't meet certain criteria. You would need to specify beforehand which people are in the white-list I guess. And automatically move all non-matching people from the "Unknown" category to the "ignored". Although, depending on how good the algorithm is, it can lead to true matches being ignored.
@ Martin Tlustos Thanks, now I understand it. As part of GSoC 2019, we will work with students to improving the facial engine. Maik
(In reply to IWBR from comment #4) > Ok, but in order to do so Digikam would still need to identify the rest of > the faces, and purposely ignore them, since it can't know if a person isn't > someone in particular until he/she has been identified. > > So what you propose is an option to hide or ignore all faces except those > that match some specific persons, right? > > In that case, it would be a step-up from what is proposed in bug #392023, in > which you could manually ignore selected faces. But in this case, > automatically if they don't meet certain criteria. You would need to specify > beforehand which people are in the white-list I guess. And automatically > move all non-matching people from the "Unknown" category to the "ignored". > > Although, depending on how good the algorithm is, it can lead to true > matches being ignored. Yes, there is a risk of true faces being ignored. But you need to weigh it against the other side - having thousands of unwanted faces you have to deal with. I don't really know what the best solution would be, I'll leave that up to you (of course... ;-)
*** Bug 363724 has been marked as a duplicate of this bug. ***
I would like to clear up my bug reports a bit. So, is this bug confirmed, assigned, fixed?
It still open... Gilles Caulier