Face Tagging

BrightBlur uses on-device face recognition to speed up the tagging process. As you use the app, it learns to recognise the people you frequently photograph.

Automatic recognition

When you upload a photo, BrightBlur compares each detected face against encrypted fingerprints of known people:

• Auto-tags — faces with high-confidence matches (above 70% similarity) are automatically assigned to the correct person. You’ll see these pre-filled during the tagging step.
• Suggestions — lower-confidence matches appear as a dropdown suggestion. Confirm with a single tap, or pick a different person.

All recognition happens on your device. The server never sees the comparison — it stores only encrypted fingerprints it cannot read.

Manual overrides

You can always override automatic decisions:

• Tag as… — choose from existing persons or create a new one inline.
• Visible to all — reveal the face to everyone in the photo’s audience.
• Blur — permanently pixelate the face. No one, including you, will ever see it again.

Missed faces

No face detector is perfect. If the automatic detection misses a face, tap Missed someone? at the end of the review flow to draw a manual bounding box. This is particularly important for faces that are partially hidden, at extreme angles, or very small in the frame.

If a photo has zero detected faces, BrightBlur shows a warning asking you to confirm there are genuinely no faces before publishing.

Correcting mistakes after publishing

If a face was tagged to the wrong person, a controller can tap the face in the photo detail view and select This isn’t [name]. This deletes the face slice and associated data — the face reverts to permanently pixelated.