Artificial intelligence algorithms are actively used for surveillance, analytics and other tasks of public control by governments and private companies in Europe and beyond. This fact reminds us of the deep oppression that has been woven into the social and technological fabric of regions and countries.
Mass surveillance affects all of us through a wide suite of technologies — but facial recognition, which has become one of the most critical and commonly-used technologies, poses special risks of disparate impact for historically marginalized communities. Facial recognition is a particularly invasive technology. It’s not only about the surveillance of activists, suspects, and minorities, but it is an invasion of privacy for everyone and an enormous danger to democratic freedoms, civil liberties, and free expression for the whole society. In the hands of private companies, police and other government agencies, face recognition technology presents an inherent threat to information security, and social justice.
Evidence shows why we need a law against biometric mass surveillance. Read here an excellent summary of extensive evidence by ReclaimYourFace (a European movement that brings people’s voices into the discussion around biometric data), documenting the rapid spread of biometric mass surveillance in EU countries.
Ban Facial Recognition Europe, this inspirational petition introduces the campaign for the permanent ban of Facial Recognition used for identification and profiling in all of Europe. An initiative by the activist Paolo Cirio in alliance with La Quadrature du Net, We Sign It, and thanks to the research of European Digital Rights (EDRi).
While the systems of surveillance continue to proliferate and permeate our lives, recops establishes a new and comprehensive methodological framework for the analysis and critique of facial recognition technologies. As a functional response to the surveillance culture, recops is a visual defense toolkit, an open-source computer vision software that challenge the ubiquitous application of facial recognition technology by governments and big tech, by using activist strategies of counter and inverse surveillance.
As a conspiracy and applied research framework; explicitly configures artificial intelligence as an act of political intervention and offers a set of tools that promotes self-reliance and therefore self-authority from bottom to top. The open-source recops software is a computer vision library, a resistance toolkit, leveraging the power of deep learning techniques and advanced visual forensic tools to assist investigative journalists, human rights researchers, and digital activists to improve their workflow for OSINT (Open-source intelligence) investigations.
Centered with facial recognition operations, recops embraces whole range of complex operations from fundamental face detection and face recognition (identification, verification or 1/1, 1/N matching) to face analysis and face clustering in images. Also, software features include advanced face appearance analysis, rich metadata extraction for each face, including age, gender, race/ethnicity, emotion classification, and more!
Aim of the project, through “transparency” and “explainability”, is to provide access, share information, and study the implications of real-world AI systems. Is pivotal step and a tactic of resistance for the continuation to the work of activists, civil society organizations and social movements. In this context recops try to feed with more accurate, efficient, and accessible tools for the collection and analysis of data to produce actionable intelligence.
The open-source recops library, is developed to be as lightweight and feature full as it could be. It is written in python. Makes heavy use of the open-source deepface library - a lightweight face recognition and facial attribute analysis library for Python; extending it’s capabilities to enable users to run deep face analysis operations. recops stores files in disk and structured data in SQL (sqlite by default), using sqlalchemy ORM. Additionally provides a simple UI using flask and flask-admin to help dealing with its content.
recops software is open-source (Do No Harm License) and available at https://github.com/forensic-toolkit/recops/. The easiest and fastest way to install the recops package is to following the guide here.