LIGHTROOM - Add Face Detection Plausibility Check Based on Date of Photo

  • 3
  • Idea
  • Updated 4 weeks ago
  • (Edited)
The face detection algorithm seems to lack a very simple parameter for the plausibility of the detection of a person: It is more likely that the same persons are on photos which are all taken at the same day.

This would mean that the face detection algorithm not just analyses each photo as an individual one, it would mean that the algorithm has to detect a whole series of photos, e.g. all photos in a given folder.

Currently the algorithm is suggesting persons from photos which are years old. I think the detection ration could significantly improved if the photos would be analysed in the context of a folder or day.
Photo of Ope Gato Cedo

Ope Gato Cedo

  • 30 Posts
  • 4 Reply Likes

Posted 4 years ago

  • 3
Photo of Jeremy Caney

Jeremy Caney

  • 24 Posts
  • 14 Reply Likes
It doesn't bother me that it finds potential matches from old photos. I do consistently find it surprising that, as you note, it doesn't (appear to) prioritize time into its probability ranking. I've had scenarios where I've done a photoshoot of a person, resulting in hundreds of photos of them from the same day, often in the same clothes, makeup, &c., but the machine learning is really slow to figure out that these are all the same people (either in terms of grouping, or in terms of suggested name).