Reverse face search

How is it different from a reverse image search?

A reverse image search compares whole images and finds copies, crops and edits of one picture. A reverse face search isolates the face, turns it into a face embedding and compares that with faces seen elsewhere, so a photo from 2015 can match a page published last week. The two answer different questions: where has this picture been published, versus where does this person appear.

What does a result contain?

• The matched image as it appears on the source page.
• The source page URL, which is what makes a result actionable: it says who published the photo and where a removal request should go.
• A rank or similarity score that says how close the two faces are, not whether they belong to the same person.

Which engines offer it?

Face-first engines include PimEyes, FaceCheck.ID, FaceSearch and Lenso.ai. General image engines such as Google Lens, TinEye and Yandex Images match images and sometimes surface faces as a side effect of visual similarity. The engines index compares them field by field. The underlying technique is one-to-many matching: one probe face against an index of many.

Where does it go wrong?

Look-alikes score high (the doppelganger effect), small or turned faces fail to match, and anything behind a login is invisible because engines only see the public web. A result is a lead to be checked against the name, date and context on the source page, never proof of identity. A missing result proves nothing either: the page may not have been crawled, or the face may be too small to index.

Where does the term come up?

In consumer privacy (finding where your own photos appear), dating safety (checking whether a profile photo appears elsewhere under another name), journalism and OSINT, and in the data-protection cases brought against engines that index faces without consent, documented under cases.