How does face search work?
You upload a photo and seconds later you get pages carrying that face. What happens in between, and how much should you trust the result?
7 min read
It searches geometry, not photos
The system first locates the face and places dozens of reference points on it: eye corners, nose tip, jaw line, brow ends.
It then measures the ratios between those points — the distance between the eyes relative to the length of the nose, and so on. Those ratios become a list of numbers, called a face signature or embedding.
What gets searched is that list of numbers, not your photo. This is why a cropped, filtered, differently lit or years-old picture can still match.
How it differs from reverse image search
Google reverse image search looks for copies of the same file. Crop it slightly and it finds nothing.
Face search looks for the person. It can find a completely different photo of them. That is both what makes it powerful and what makes it worth taking seriously.
What the scores mean
Results come with a similarity score from 0 to 100. Roughly: above 85 is near-certain, 70 to 85 is likely, below that is weak.
But it is a measure of likeness, not proof of identity. Siblings, relatives and occasionally unrelated people score highly. With identical twins the system cannot meaningfully tell them apart.
Why it sometimes finds nothing
The index is not the whole internet. Private accounts, pages behind logins and unindexed sites are all outside it.
Photo quality matters too: low resolution, side angles, sunglasses, masks, heavy filters and very old photos all reduce the hit rate.
No result does not mean the person is not online. It means they were not found in indexed sources.
The legal side
Face data is biometric data, protected far more tightly than ordinary personal data — special category data under GDPR Article 9, with equivalents in Turkey's KVKK and Illinois' BIPA.
In practice: searching your own face, searching with someone's permission, or verifying who you are talking to are legitimate uses. Stalking someone, or using results for hiring, housing or credit decisions, is not.
Common questions
- What is the difference between face search and reverse image search?
- Reverse image search looks for copies of the same file. Face search looks for facial geometry, so it can find a different photo of the same person.
- What match score is reliable?
- Above 85 is considered strong and 70 to 85 likely. Even so, the score measures similarity rather than identity — relatives and lookalikes can score highly.
- Why do some searches return nothing?
- The index does not cover the whole internet; private accounts and pages behind logins are excluded. Low resolution or side-angle photos also reduce accuracy.
- Is face search legal?
- It depends on the use. Searching your own face or verifying someone you are talking to is legitimate; stalking, or using results for hiring and similar decisions, is not.
Now you know how it works — you can try it with your own photo.