[Feature][faces] try to detect a person even if their face is not visible #20958
Closed
ireun
started this conversation in
Feature Request
Replies: 3 comments
|
cc @mertalev, thoughts? FWIW
They've got hundreds of people dedicated to figuring that kind of thing out ;) |
0 replies
|
I'm not aware of a model that recognizes a person by their body or clothes rather than by face. If this exists, it could be a good addition. My concern is whether this would consistently be an improvement, or if it could cause clustering mishaps in some cases. |
0 replies
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
I have searched the existing feature requests, both open and closed, to make sure this is not a duplicate request.
The feature
Hi!
Maybe I'll give an example as this is pretty hard to explain:

Immich is able to detect a face photos 2, 3 and 4 (buffalo_l, 0.6, 0.3, 3)
It is pretty clear (for a human) that person in photo 1 is the same one.
Would it be possible to detect a person, rather then just face? (Human Pose Estimation? https://docs.ultralytics.com/tasks/pose/)
If yes: Would it be possible to search for a similar person (clothes, hair) in the nearby photos (time, place, exif-data) and match people through that? (even if it would need manual confirmation) (https://github.com/KaiyangZhou/deep-person-reid ?)
The app-that-can't-be-names is able to achieve this.
Platform
All reactions