People & Face Grouping

Fovea uses 100% on-device AI to detect faces and cluster them by identity — no cloud, no uploads, fully private.

How It Works

When you import photos, Fovea’s AI scans every image for faces using Apple’s Vision framework.
Detected faces are encoded into embeddings using a neural face encoder.
The Hierarchical Clusterer groups these embeddings, identifying which faces belong to the same person.
Results appear in the People sidebar section.

Using the People View

Navigate to the People section from the sidebar. You’ll see:

A grid of person clusters, each represented by a circular crop of their face from one of their photos.
The number of photos each person appears in.
Click any person to see all their photos in a detail view.

Privacy

All facial recognition and clustering happens locally on your Mac. Face data is never sent to a server. Fovea doesn’t require an internet connection for AI features.

Tips

Face grouping works best with clear, well-lit portraits. Heavily obscured or profile-only faces may not be clustered reliably.
The more photos you import, the more accurately the clustering algorithm can group people together.