Skip to main content
gregarmoured
Participant
March 6, 2026
Open for Voting

P: Face recognition - allow user to specify criteria (eg capture time) to exclude names from suggested matches

  • March 6, 2026
  • 0 replies
  • 11 views

Feature: allow users to provide criteria for name-space search (face-name matches) that exclude (prefer) according to user-specified criteria. Most important criteria for matching is capture time, but other criteria (keyword / keyword groups, for example) could be used. Fuzzy logic implementation would also be possible (e.g. heavily weight likely results by year of capture, presence in collections/collection sets, etc)
This will GREATLY reduce the number of false-positive matches.
Example using capture time:
-User has large catalog containing many photographs, including images going many years.
-Selects a group of photos (by collection set, folder years, search criteria, etc) that all have capture times of 2025.
-In suggested matches, Lightroom currently displays potential matches for names that are only in images prior to 2005 (real world example). 
-Provide an option for user to specify capture-time periods (eg 2020-present), and the number of false matches drops dramatically. 
-All of this data already exists in the Lightroom catalog internal database, and implementation could be done in background (i.e. present possible matches but remove those suggestions that do not fit criteria as they are identified, in same way that the people panel view moves/removes potential matches as analysis proceeds.
-For some criteria, fuzzy logic (non-exact matching) would also improve results (eg allow names from slightly outside the specified criteria, as long as 'proximate' in time or in other collections within a collection set (tree hierarchy could be used to weight likelihood/exclusion).
-Similarly, the same approach could be used for other criteria, such as keywords (eg 'family', 'hockey', 'work', 'college' whatever), presence in collection sets, or other.

Note, while I am suggesting a user-controlled mechanism to define criteria, nothing presents using this approach to 'weight' results in the people view suggestions (Lightroom may already be doing this to some degree but weighting criteria does not seem to play much of a role, despite it being an ‘AI’ feature, it is clearly not making much use of eg capture time).