Lightroom Classic: Known Unknown People

  • 3
  • Idea
  • Updated 2 months ago
  • (Edited)
When going through facial recognition, there are often people whose name is not known—and may not be relevant to find out. For instance, if I'm photographing a wedding, there may be ancillary guests that show up repeatedly. Because I can't name them, they continue to show up in the "Unnamed People" category, and can make it difficult to continue tagging.

Ideally, I'd be able to flag these people as an "Unknown Person"—or something like that. Basically, a way of telling Lightroom, "Yes, I recognize this is a person, and I have no intention of naming them". This way they could be moved out of "Unnamed People" even though they don't have a known name.

Obviously, I could instead just delete the Face Region. And, in fact, that's what I used to do. But this can be problematic if, at a later date, I decide the person is relevant, and would like to go back and name them. 
Photo of Jeremy Caney

Jeremy Caney

  • 26 Posts
  • 14 Reply Likes

Posted 2 months ago

  • 3
Photo of Ronald Chambers

Ronald Chambers

  • 68 Posts
  • 11 Reply Likes
Why don't you name the unknown person as Unknown Person0000001 or something like that? Other times they show up they should be found under that name. When you do find out who they are you can correct the name and proceed.
RONC
Photo of Jeremy Caney

Jeremy Caney

  • 26 Posts
  • 14 Reply Likes
Hi Ronald, there are several reasons that's not a workable solution:
  1. I'd need to uniquely identify each unknown person by typing in a fictitious name (and likely, as in your example, keeping track of my index)
  2. If I did want to group "Unknown People" (which I would) then I would need to keep track of what number I assigned that person
  3. Those people would get mixed in with "Named People", even though they're otherwise not significant

I do actually do this for people who a) I can come up with a memorable name for, and b) who show up repeatedly across facial recognition (e.g., "Yellow Jacket" for the guy wearing a bright yellow rain coat at a wedding). But that's not a scalable approach.
Photo of Aidan Wojtas

Aidan Wojtas

  • 64 Posts
  • 24 Reply Likes
1. zUnknownPerson_(DateTime that you named them) e.g. zUnknownPerson_20190919_1655
2. No, see above it would be unique
3. Name with "z" at start, will be at the end and less confusing.
(Edited)
Photo of Jeremy Caney

Jeremy Caney

  • 26 Posts
  • 14 Reply Likes
Yes, fair enough regarding using the date and a “z” prefix to ensure uniqueness and sorting.

But the grouping is highly desirable so that e.g. one person that shows up in a hundred photographs doesn’t take up a hundred slots. And, regardless, nobody is going to go through and uniquely name every face in a crowd—a wedding, for instance, might have hundreds of people, with some showing up in hundreds of photographs.

I get that you’re just suggesting a workaround given the current tools available, and I sincerely appreciate that. But there’s a relatively simple and elegant approach to baking this into the product, which is why I’ve suggested this feature. It would save time, keep the People section organized, and also help train the machine learning algorithm so that it’s not continually re-evaluating each against named people (which is what happens today).
Photo of Roelof Moorlag

Roelof Moorlag

  • 210 Posts
  • 62 Reply Likes
As Aidan described it maybe can be done buth it's not very 'workable'. Jeremy's suggestion would be a improvement for the whole face recognition system in Classic and i welcome every improvement of it. You got my vote.
Photo of Jeremy Caney

Jeremy Caney

  • 26 Posts
  • 14 Reply Likes
It's worth noting that I see this idea as closely related to another one, proposed four years ago:

LIGHTROOM - Group unnamed person

Either of these could be implemented independently, but they would work very nicely together. Both are acting to declutter the "Unnamed People" category, while acknowledging that it's not always practical to provide accurate or even predictable names to every face.