Face Recognition

Automatically detect, cluster, and organize faces in your photo library with production-grade computer vision. Built for professional workflows where accuracy and review control matter.

How It Works

1. Face Detection

When you upload photos, advanced face detection identifies all faces in each image. The system extracts:

• Face location and bounding box
• Confidence score (0-100)
• Quality indicators
• Face pose angles

2. Embedding Generation

Each detected face is converted into a 512-dimensional mathematical representation (embedding). These embeddings capture the unique characteristics of each face.

3. Automatic Clustering

Using graph-based clustering, the system groups similar faces together automatically. No manual sorting required - faces are organized by similarity.

4. Review & Confirmation

Review the generated clusters in the sidebar. Each cluster shows:

• Representative face photo
• Number of faces in cluster
• Average confidence score

Confirm each cluster and assign a project tag. Once confirmed, all photos in that cluster are tagged.

5. Auto-Tagging New Photos

When new photos are uploaded, faces are automatically matched against confirmed clusters. High-confidence matches are auto-tagged. Lower confidence matches are queued for review.

Quality Analysis

The system automatically detects quality indicators:

• Sunglasses: Face partially obscured
• Profile: Side angle rather than frontal
• Poor Lighting: Exposure or contrast issues
• Small Face: Face too small for reliable detection
• Occlusion: Face partially blocked

These indicators help prioritize which photos to use and which may need manual review.

Cluster Management

Confirm Clusters

Review unconfirmed clusters and assign tags. Once confirmed, tags are applied to all photos in the cluster.

Ignore Clusters

Mark clusters for unimportant people (background extras, crew) to keep your focus on key talent.

Remove Faces

Remove incorrectly grouped faces from a cluster before confirmation for accurate tagging.

Reset Clusters

Reset individual clusters or all project clusters to re-process with updated photos.

Technical Details

• Background Processing: Detection and clustering run asynchronously
• Progress Tracking: Real-time status updates during processing
• Scalable: Handles 50,000+ distinct individuals across millions of images
• Per-Project Isolation: Face data is isolated to each project

Privacy & Security

• Face embeddings are mathematical representations, not searchable images
• All data stored encrypted in your project
• Processing happens on secure infrastructure
• Only you control access to your tagged photos

Use Cases

Production Stills

Tag entire cast across thousands of unit stills. Review clusters once, and let auto-tagging handle the rest as new photos arrive.

Event Photography

Organize photos by attendees. Confirm VIP clusters first for priority delivery.

Sports Photography

Group athletes across game photos. Build individual portfolios with consistent tagging.

Perfect For

Unit still photographers, event photographers, and production teams managing large photo libraries where consistent face tagging is essential for organization and delivery.