Skip to main content

People

The People tab groups faces across your library so you can name family members and browse photos by person. Eyedeea uses AI to detect faces and group similar faces into clusters - doing most of the matching work so you only confirm who is who.

On this page


How it works

  1. After Auto Curation runs, the AI detects faces in your photos.
  2. Similar faces are clustered - often the same person across many albums.
  3. You review clusters: name them, merge duplicates, or mark mistakes.
  4. Over time, recognition improves as you confirm identities.
People tab showing named face clusters

Review face clusters

The People grid shows:

  • Named people - clusters you have already identified, with a face thumbnail and photo count
  • Unidentified clusters - labeled Who's this? until you assign a name

Tap any card to open the identity editor. Use the search box at the top to find a person by name.

Unidentified Who's this face cluster card - screenshot pending

Name a person

Web and Android:

  1. Tap a face cluster (named or unidentified).
  2. Type a new name or pick an existing person from the suggestions list.
  3. Tap Save.

You can leave clusters unnamed - they still get stable IDs for later review.

People identity editor with name field and suggestions - screenshot pending

Not a face and Do not track

Sometimes the AI detects a non-face (a pattern, poster, or statue) or someone you prefer not to track.

ActionWhen to use
Not a faceThe detection is wrong - not a real face
Do not trackA real face you do not want grouped or named
Not a face and Do not track buttons in People editor - screenshot pending

On mobile, help text for these actions may appear in a collapsible details section.


Correct a wrong face

When a face is tagged with the wrong name (for example Deb’s face appears under Robert):

  1. Open the named person from the People grid (merge / detail panel).
  2. Tap Review faces.
  3. Select the face crops that do not belong to that person.
  4. Tap Move to… and choose the correct person, or type a new name.
  5. Optionally use Not a face or Do not track for bad detections instead of moving them.

If moving faces leaves a person with no faces left, you can remove that empty person when prompted.


Merge people

When the AI split one person into two clusters (common after hairstyle changes or years apart):

  1. Tap a named person to open their detail panel.
  2. Review stats - photos, faces, clusters, and aliases.
  3. Choose Merge person and pick the duplicate identity to combine.
  4. Confirm in the merge dialog - this cannot be undone easily.
Merge people Icon
People merge side panel with stats and merge button

Browse photos by person

Web:

  • Tap a named person's photo count or face thumbnail to open Search filtered to that person.
  • From search results, open any photo in the Photo Viewer.

Android app:

  • The dedicated Search tab is hidden on Android. Person-based browsing works best through Player Settings search, or use the web browser on a laptop for the full Search tab experience.

Source selector

Households with multiple photo sources may see a source selector at the top of People. Pick a source to filter which library the face grid shows.

People tab source selector dropdown - screenshot pending

Web and Android

The People UI is the same on web and Android. The main difference is how you search by person afterward - web has a dedicated Search tab; Android relies on Player Settings.


Privacy

Face data stays inside your household. Only members with library access can see People albums. Face analysis runs on your library to power People browsing - it is not shared outside your household account. Eyedeea does not sell or share face data with third parties.


Good to know

  • People counts also appear on Summary Action Center when unnamed clusters need review.
  • Removing a person from a single photo happens in the Photo Info panel inside the Photo Viewer - that does not delete the person from your library.
  • For sensitive photos that include faces, review flagged items in Sensitive separately.