Hi @luator,
I can help answer your questions.
- What exactly happens under the hood when I remove a face vs. marking it as ignored? I guess when removing it, it may later be re-detected, but is the information also used to refine the detection (i.e. ignoring is confirming that this is indeed a face while removing means “this was a false positive”)? In other words, does it harm the future detection quality if I remove a face that is actually a face (just not someone I want to tag)?
Removing a face suggestion has no long-term effect. We don’t save the rejection, and the rejection isn’t used to train the recognition engine. Ignoring the face marks the face so it is not used in any future recognition scans.
- How to best handle images with lots of unknown people in it. Let’s say a photo from a public event, where ~20 faces are detected but I don’t know any of them. Is there a way quick way to remove/ignore all faces of an image (so I don’t have to click “ignore” 20 times)?
Yes! If you have the full image open (not just a thumbnail), you can right-click on the image and click “ignore all unconfirmed faces” in the context menu.
- Related to the question above: Is it possible to somehow exclude an album from automatic face detection? Again the use case would be photos from an event where there are lots of faces but 99% of them are strangers which I don’t want to tag.
Yes again, but it’s a little tricky. The middle tab in the face scan settings allow your filter in or out specific folders.
Let me know if you have any questions.
Cheers,
Mike
Source: I rewrote a considerable portion of the face engine in digiKam 8.5.0.