Going through the face detection and recognition workflow of over 20,000 photos. I've pushed the accuracy up to 95% to start and it has been reliable, but as I pull back to 90%, there are a lot of misses.

As a result, I find myself going into unknown faces and tagging them there. It will be very useful if faces that were 99% (for example) similar to be grouped together. So when I assign that one face to the tag, all of it's other similar faces would be assigned the same tag.

So rather than doing 1 face at a time. You can imagine how much quicker assign tags to 3-20 faces at a time can be.

A previous ticket seems to have a similar thought but I don't see any mechanism to group similar faces as it describes.
https://bugs.kde.org/show_bug.cgi?id=392024