How to set up AI face matching for a photo day
AI face matching is what makes a 5,000-subject school district workable. Done right, every parent receives a magic-link gallery containing only their child's photos — no proofing, no manual tagging. The trick is capturing reference photos at the start of the day.
1.Confirm the roster is imported and locked
AI face matching keys off the roster. Before photo day, verify that the roster import is complete and every subject you expect to photograph has a row. Mismatches between roster and reality at the venue are the most common cause of bad matches.
2.Install the Reference Capture App on your Android device
Open the Pro Photo Systems Android Reference Capture App on the phone or tablet you'll use at the venue. Sign in with your studio account, then select the event. The app stores roster data locally so it works without Wi-Fi at the venue.
3.At the venue: confirm subject identity, then capture
For each subject, search the roster in the app (by name or group), confirm the right person is in front of the camera, then tap to capture a single reference photo. The app stamps the photo with the subject's identity and queues it for upload. One good reference photo per subject is enough.
4.Continue shooting production photos
While reference capture is happening (often by a helper at the front of the line), the photographer captures the actual production photos. These are uploaded later from the camera's SD card to Pro Photo Systems, not via the Capture App.
5.Upload reference photos when you have Wi-Fi
The Capture App auto-syncs reference photos when it has connectivity. You can leave the venue with all reference photos queued; they'll finish uploading once the device is on a working network.
6.Upload production photos to the event
Back at the studio, upload your full production set to the event in Pro Photo Systems. AI face matching runs automatically — every production photo is matched against the reference photos for the event, and the system writes a confidence score for each match.
7.Review matches and approve delivery
Review any matches below your confidence threshold (configurable per event). High-confidence matches auto-deliver to the right parent's gallery; low-confidence ones get queued for human review. When you're satisfied, send the magic-link emails to parents.