Face recognition for event photos: how guests find their own pictures

How it works, in four steps

1. Photos are uploaded. The photographer’s camera sends photos to the cloud during the event.
2. Faces are indexed. Each face found in a photo is turned into a numeric description, a “search vector”, stored against that photo. This is not a picture of the face.
3. A guest takes a selfie. After scanning the event QR code and agreeing to the prompt, the guest takes one selfie, which is converted into a search vector in the same way.
4. The two are compared. Photos whose faces are close to the selfie’s vector are shown to the guest as their matches.

The guest never has to scroll through the event’s full set of photos to find themselves.

What happens to the selfie and the photos?

This depends entirely on the provider, so it is worth checking. For SnapTrace, the published answers are:

• The selfie is not kept. It is used for one search and then discarded.
• No personal details are collected from guests. No name, email, phone number or account is needed.
• Matching is limited to one event. A selfie is only compared with the photos from the event whose QR code the guest scanned.
• Consent comes first. A guest agrees to face matching before any search happens, and guests under 18 need a parent or guardian to give consent and take the selfie with them.
• Everything is deleted. Event photos, face index data and guest search history are deleted automatically 5 to 14 days after the final upload, depending on the photographer’s plan.
• The photographer controls downloads. They choose whether guests can only view photos or also save them.

The full wording is in our Guest Privacy Notice.

What affects accuracy

Face matching works well on clear, front-on and slightly angled faces, and it is more challenging with strong shadows, profile views, sunglasses, heavy headwear or faces that are very small in a wide shot. SnapTrace’s matching runs on AWS Rekognition, an established face-analysis service from Amazon Web Services, but no system is perfect. A guest may occasionally miss a photo or see one that is not theirs. Photographers can also make a highlight album available to everyone.

Questions to ask any provider

• Is my selfie stored, and for how long?
• What do you need from guests: a name, an email, a phone number, an account?
• Is consent asked before the face search, and what happens for children?
• Is the selfie compared only with this event’s photos?
• When are the photos and face data deleted, and can I ask for my match record to be removed sooner?
• Are faces ever used to train AI models or sold?

For SnapTrace the last answer is no: guests’ data is not sold and faces are not used to train AI.

If you are the photographer

Face matching is what turns a gallery of thousands into a personal one for every guest, without you sending a single link. See how to deliver event photos the same night and SnapTrace pricing, or talk to us.

Guest reading this? How to share wedding photos with guests compares the other options.

More from SnapTrace

SnapTrace is a face recognition photo delivery platform for event photographers in India, built by UNIT3A. See how it works, or browse all guides.