Facebook Tag Suggestions

Facebook's Tag Suggestions feature used facial recognition to scan faces in users’ photos and offered suggestions about who the person might be. Introduced in June 2011, the feature was enabled by default and ran on virtually every face in photos uploaded to the Facebook platform.

Per EPIC, Tag Suggestions worked through a four-step process in which the tool tries to detect faces in uploaded images; standardises, or “aligns,” the face along a set of parameters, such as orientation and size; computes a “face signature,” which is a string of numbers that represents that particular face; and searches a database of stored “face templates” for a match.

The stored face templates are calculated based on other photographs that a user is tagged in. A match occurs when the face signature falls within a threshold of similarity to a stored face template, at which point Facebook suggests tagging the user to whom the face template is assigned. Facebook claimed that they only store face templates, and not face signatures. 

In 2019, Facebook discontinued its use of Tag Suggestions, replacing it with its general facial recognition system which is turned off by default and used for a variety of purposes, including the identification of new users.