In an album with 800 photos it is hard to find the nine where grandma appears. Face grouping collects photos of the same person so you can find them in a second – without the photos leaving your album.
When a photo is uploaded it is analysed in the background on our servers – not on the guest's phone. That means the album is not rescanned every time someone opens it, and no phone battery is spent on it. The result is stored with the album.
As the owner you can name a face group, for example 'Grandma'. After that, every photo with that person is one click away. If two groups look alike they can be merged, and wrong suggestions can be rejected.
If a person is turned away or partly hidden, you can mark the area on the photo manually and type a name. That is useful for group photos and for children, which automatic detection finds harder.
Face data belongs to the individual album and is used only to group photos inside it. It is not shared across albums, not sold and not used to train models. Delete the album and it goes with it.
At work, mention the feature before the party and respect it if an employee does not want to be named. Naming is always an active choice by the host – nobody is labelled automatically.
No. The analysis runs on our servers when the photo is uploaded, so the album never has to be rescanned when you open it.
No. The system groups similar faces. Names are only added when the host actively does it.
No. It belongs to the individual album and is deleted along with it.
Yes. You can merge groups, reject suggestions and tag people manually on the photo.