Facial Recognition: Classification

As noted by the French Data Protection Authority (DPA), “Commission Nationale de l’Informatique et des Libertés” (CNIL) in November 2019, “the current debate on facial recognition is sometimes distorted by a poor grasp of this technology and how it exactly works”. The objective of the present paper is, precisely, to present, in the most accessible way possible, exactly how facial recognition and facial analysis work.
- from “Facial Recognition: Classification.” Theodore Christakis, Professor of International and European Law at Université Grenoble Alpes, is the project leader.

Facial recognition technologies (FRT) perform functions such as authentication, identification, surveillance, and emotion recognition. They involve processing very sensitive biometric data. Unfortunately, debates over how to regulate the use of FRT lack clarity, consistent definitions, and a comprehensive understanding of how the distinct functionalities work.

The MAPping the Use of Facial Recognition in Public Spaces in Europe (MAPFRE) project is an independent study with the objective of analyzing the different functionalities and uses of facial recognition, exploring the related legal issues (i.e., data protection, algorithmic bias, use of facial analysis within the context of criminal investigations), and presenting 25 use cases.

In “Facial Recognition: Classification,” the second of six reports from the MAPFRE project, Theodore Christakis (project leader) and his coauthors Karine Bannelier, Claude Castelluccia, and Daniel Le Métayer provide a path to understanding how the different facial recognition and facial analysis technologies work. This report includes a “Classification Table” which details how the different facial processing functionalities and applications are used in public spaces. Below is the Executive Summary from this first report.

Executive Summary from “Classification.”
This is the second report from the Mapping the Use of Facial Recognition in Public Spaces in Europe (MAPFRE) project. It is written by Theodore Christakis (project leader), Karine Bannelier, Claude Castelluccia, and Daniel Le Métayer. Additional contributors are: Alexandre Lodie, Stephanie Celis Juarez, Coralie Pison-Hindawi, and Anaïs Trotry.

In Part 1 of our “MAPping the use of Facial Recognition in public spaces in Europe” (MAPFRE) project we explained in detail what “facial recognition” means, addressed the issues surrounding definitions, presented the political landscape and set out the exact material and geographical scope of the study. Furthermore, we explained how our study covers all the ways in which face processing systems are used in public spaces in Europe, whether the data involved are “biometric data” or, to use the new term, are “biometrics-based data”. Drawing on the draft EU AI Regulation, we also precisely defined what we mean by the term “public spaces” and presented three subcategories, that we have used for our study: “open spaces”; “restricted spaces”; “closed spaces”.

As noted by the French Data Protection Authority, CNIL [Commission Nationale de l'Informatique et des Libertés], “the current debate on facial recognition is sometimes distorted by a poor grasp of this technology and how it exactly works”. The specific objective of the present paper is to present how facial recognition and facial analysis work.

We have also endeavoured to produce a “Classification Table” detailing how facial recognition/ analysis is used in public spaces. This classification table tries to present in the most accurate and accessible way the different facial processing functionalities and applications used in public spaces, which encompass the various forms of both “face recognition” and “face analysis”.

We hope that this classification table, together with the illustrations, explanations and numerous examples that are included, which are based on our “25 selected case studies”, will serve as a useful tool in preventing the various uses of facial recognition being conflated, and will bring further nuance and clarity to the public debate.

Classification Table
Classifying the uses of facial recognition/analysis in public space

Read the full report, Part 2 of Mapping the Use of Facial Recognition in Public Spaces in Europe (MAPFRE): “Classification” from the AI-Regulation.com website.

Part 1 of the MAPFRE project, “A Quest for Clarity: Unpicking the 'Catch-All' Term” presents the current political landscape, dives into an analysis of the problems of definitions for key facial recognition terms, and explains the project’s main objectives and methodological tools.

Part 3 of the MAPFRE project, “Facial Recognition for Authorisation Purposes” is the first ever detailed analysis of what is the most widespread way in which Facial Recognition is used in public (& private) spaces: to authorise access to a place or to a service.

Parts 4 through 6 will be published in the near future. Look for them on the article page of the AI-Regulation.com website.

This Executive Summary of the second report from the Mapping the Use of Facial Recognition in Public Spaces in Europe project, titled, “Classification” was first published on the AI-Regulation.com website on May 17, 2022. It is reproduced here with the kind permission of the project leader, Professor Théodore Christakis.

Disclosure: Microsoft is a corporate sponsor of AI-Regulation.com, the website of the Chair on the Legal and Regulatory Implications of Artificial Intelligence at MIAI Grenoble Alpes, and Microsoft also sponsors the Technology | Academics | Policy (TAP) website. Microsoft respects academic freedom, and is working to enable the dialogue on the most critical tech policy issues being debated.