Face Recognition

Tagging 3,000 unit stills by hand takes a full day you don't have. Face recognition does the sorting for you, so you spend minutes reviewing instead of hours clicking.

How it works

1. Face detection

When you upload photos, the system scans every image and identifies faces. For each one, it records:

• Face location and bounding box
• Confidence score (0-100)
• Quality indicators (lighting, angle, obstructions)
• Pose angles

2. Building face signatures

Each detected face gets a unique digital signature that captures its distinguishing features. You don't need to understand the math — just know that two photos of the same person will produce very similar signatures, even across different lighting, angles, and expressions.

3. Automatic clustering

The system groups similar face signatures together. You don't sort anything manually. Faces that look alike end up in the same cluster.

4. Review and confirmation

Open the cluster sidebar and you'll see each group with:

• A representative face photo
• How many faces are in the cluster
• The average confidence score

Confirm a cluster, assign a project tag, and every photo in that group gets tagged instantly.

5. Auto-tagging new uploads

Photos you upload later get matched against your confirmed clusters. High-confidence matches are tagged automatically. Lower-confidence ones go into your review queue.

Quality analysis

The system flags potential issues so you can prioritize your review time:

• Sunglasses: Face partially hidden
• Profile angle: Side view rather than frontal
• Poor lighting: Exposure or contrast problems
• Small face: Too small for reliable detection
• Occlusion: Face partially blocked by another object or person

These aren't rejections. They're signals that help you decide which photos to check first.

Cluster management

Confirm clusters

Review unconfirmed clusters, assign tags, and all photos in that cluster get tagged at once.

Ignore clusters

Background extras, crew members, bystanders. Mark their clusters as ignored so you can focus on the talent that matters.

Remove faces

Spot a face that landed in the wrong cluster? Pull it out before you confirm. Keeps your tags accurate from the start.

Reset clusters

Reset individual clusters or the whole project. Handy when you've added a big batch of new content and want the system to re-process everything.

Technical details

• Background processing: Detection and clustering run while you keep working
• Progress tracking: Status updates while processing runs
• Project isolation: Face data stays within each project. Nothing crosses over

Privacy and security

Face signatures are not searchable images of people. All data is stored encrypted within your project. Processing happens on secure infrastructure, and only you control who sees your tagged photos. We don't share or sell your data.

Who this is for

Unit still photographers dealing with thousands of shots per production. Event photographers organizing by attendee. Production coordinators who need consistent tagging across a project without burning hours on manual work.