Manage People by Naming, Merging and Correcting Face Groups
Turn unnamed face groups into a reusable people list, review assigned faces, correct mistakes and distinguish photo counts from face counts.
Turn unnamed face groups into a reusable people list, review assigned faces, correct mistakes and distinguish photo counts from face counts.
Person management answers “Who is in this library?” and “Which photos belong to this person?” Face search finds candidates for one reference face; person management maintains named groups over time. Automatic clustering provides a starting point that you review and correct.
1. Open the People panel
Switch to People on the left of the search workspace. The panel contains unnamed groups, named people and ignored faces.
Review each group before naming it. Ignore irrelevant detections or faces you do not need to manage, rather than mixing them with confirmed people.
2. Name a group
Use Add Name beneath an unnamed group after confirming that its faces belong together. The person then appears in the named list.
When another unnamed group belongs to an existing person, assign it to that name. Existing-person suggestions can help avoid duplicate names. Use stable names or project identifiers that your team recognizes.
3. Browse a person's photos
Select a named person to view associated photographs. Open previews and use the page actions for managing faces, exporting, and refreshing.
If the count looks too small or too large, inspect unnamed groups and the assigned faces before assuming the person was fully recognized.
4. Correct assignments with Manage Faces
Open Manage Faces on a person's page to inspect face IDs, photo IDs, dimensions and face boxes, and use actions such as copying a path.
Remove a wrongly assigned face from the person to return it to the unnamed area. This does not delete the original photograph. Review similar-looking people, profiles, obscured faces, strong lighting and low-resolution detections carefully.
5. Rename, merge, delete or ignore
- Rename a person to correct spelling or standardize your naming scheme.
- Merge entries when one person has been split across several names.
- Delete a person grouping to return its faces to the unnamed area; this does not delete source photos.
- Restore ignored faces if you need to organize them later.
Removing or deleting an actual image is a separate action in the file or image menu.
6. Choose the right workflow
Use Face Search for a one-off reference lookup, People for a maintained named index, and Face Album for grouping and reviewing a set of photographs. After parsing faces, name and correct groups, then use search, collections and export to prepare a selection.
7. Preparation and limits
The library needs face detections and features. If the relevant feature is unavailable, check parsing status and complete the face-data repair where needed.
Clustering is not a final identity determination. Profiles, masks, glasses, makeup, changes over time, group photos and low resolution can affect results. Name clear groups first, leave uncertain ones unnamed and inspect full images before delivery.
Photo counts, face counts and RAW export
From 3.13.29, people lists distinguish photos from faces. A photograph can contain more than one face assigned to a person, so these counts—and the number of exported photos—need not match.
RAW export from People, Face Album and image clustering copies the original RAW file rather than its preview. Check that the source is accessible and the destination has enough space.
Worked Example
Two copies of one public sample portrait form one unnamed group. Naming it 示例人物 creates a person with two photos. Removing one face through Manage Faces and confirming returns it to Unnamed: the named person has one photo and Unnamed has one face. The original files remain. The screenshots show this small workflow; identical copies do not demonstrate recognition across different portraits.
Sample image: NASA public-domain astronaut image distributed in the scikit-image sample dataset.