Portrait Material Management: From Indexing and Face Grouping to People Search and Export
For event, client, employee, or portrait libraries, generate face data, use face search and face albums, set scopes, review results, and export photos by person.
For event, client, employee, or portrait libraries, generate face data, use face search and face albums, set scopes, review results, and export photos by person.
Portrait workflows need face data and careful review. This guide covers the full route from indexing to export.
1. Prepare Face Data During Import
When adding portrait folders, enable Face Search and click Start Adding if you plan to search by person or use face albums.
Start with one project folder, then open the task indicator and wait for detection and feature extraction to complete. Partial data can produce incomplete searches and groups.
2. Start a Search with a Clear Face
Use a clear reference face. Crop if the face is small or surrounded by other people.
Open Face in the toolbar, upload or paste a reference, wait for detection and select the target face box if several people appear. Set the similarity threshold and result count before starting.
3. Control Face Search Scope
Search the whole database when the person's folder is unknown. Search a specific folder or collection when the event or project is known.
4. Review Face Search Results
Check whether the matched person is truly the same. Similar faces can appear in large portrait libraries.
Profiles, occlusion, makeup and lighting changes can affect scores. Open full previews and confirm identity and event context before delivery.
5. Use Face Filter for Attribute-based Shortlists
Without a reference face, use Face Filter to select face count, estimated age, smile, glasses, beard or the face's share of the image. Choose a database, folder or collection scope. This filters attributes; it does not identify a person.
An older library may have face boxes and features but lack attributes. Complete Repair Database → Face Database before using those conditions.
6. Use Face Album for Person-based Grouping
Open the face album to browse detected person groups and review photos by person.
Use More → Face Clustering, select the event folder and summarize. Adjust the album's similarity threshold to change grouping strictness. For maintained names, merges and corrections, use the separate People panel.
7. Adjust Face Settings
Settings → Face controls large-image detection and tiling. Use it when small faces in large photographs are missed, restart as required and reprocess a sample. These detection settings are separate from Face Album's grouping-similarity threshold. Changing a setting does not automatically recompute old face data. Keep a database backup before reprocessing and review maintained person assignments afterward.
8. Export Face Grouping Results
After review, use the top Export action to export all current groups. To deliver one event, use Folder Filter to narrow the source and summarize again before exporting.
Confirm the destination, available space and source access. Exported group numbers are not automatically verified real names. For RAW material, export preserves the original RAW file rather than substituting the preview.
Common Cases
- Face not detected: use a clearer image or crop the face area.
- Wrong matches: narrow scope and review manually.
- Grouping is too broad or split: adjust the album threshold and summarize again. For missing faces, check source access and detection data before reprocessing.
Screenshot Example
The NASA public-domain portrait from the scikit-image sample dataset and its identical backup form one group containing two photos at threshold 40. These screenshots show the same sample used in Face Search. Identical copies illustrate the workflow, not recognition accuracy across different portraits; 40 is the example setting, not a threshold recommendation for every event.