The Algorithmic Security Politics of Computer Vision

Projektinformationen

SECURITY VISION

ID Finanzhilfevereinbarung: 866535

DOI

10.3030/866535

EK-Unterschriftsdatum 8 April 2020

Startdatum 1 Januar 2021

Enddatum 30 Juni 2026

Finanziert unter

• EXCELLENT SCIENCE - European Research Council (ERC)

Gesamtkosten

€ 1 998 320,00

EU-Beitrag

€ 1 998 320,00

Koordiniert durch

UNIVERSITEIT LEIDEN
Netherlands

Periodic Reporting for period 3 - SECURITY VISION (The Algorithmic Security Politics of Computer Vision)

Berichtszeitraum: 2024-01-01 bis 2025-06-30

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

SECURITY VISION explores the impact of computer vision on security, with a focus on the theoretical, empirical, and political implications of this technology. The project aims to document the workings of machine vision as a human-machine interface, understand how technological and social factors explain the different uses of the technologies, and determine how these various deployments are impacted by the routinized patterns of bureaucratic labor and bureaucratic politics that empower and marginalize actors in the field.

The project proposes a nested set of methodological approaches to reach this objective: First, a global mapping of computer vision technologies in the field of security. Second, based on the first step, innovative multi-modal research design which uses visual ethnography, and critical coding to dive in-depth in specific cases: crowd control and smart technologies in urban settings such as streets and stadiums, emotion recognition used for lie detection in judicial investigation or at the border, and gait/movement recognition in the context of violence prevention and surveillance. The project also explores the ethical and political impact of computer vision, taking a comparative view across the three cases.

On the basis of the detailed ethnographic fieldwork of the researchers, and in collaboration with them, SECURITY VISION analyzes how the micro-politics of attributing suspicion and guilt are distributed throughout human-machine interactions and routines. The project aims to contribute to a better understanding of the implications of computer vision for security, and to inform policy and practice in this field

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

At the collective level, the project has almost completed the initial mapping phase, which took much more time than anticipated, but also yielded more results. We chose two main mapping routes. The first involved creating a quali-quantitative database on biometric mass surveillance. This comprehensive database is the foundation for at least two visual projects; an interactive map and report presented at the European Parliament in 2021, and a set of experimental maps under development. Several publications came out of this process, including open-source code and several peer-review articles on theory, methods, and politics of data visualization. Simultaneously, the second route explored computer vision in security through the innovative method of time-based diagramming, which led to the creation of a new software tool for time-based recording and resulted in two peer-reviewed publications. Finally, team members pursued their individual multimodal projects. PhD student Ruben van de Ven began fieldwork on movement recognition in the Netherlands through critical coding methods. Cyan Bae is starting her visual ethnography of emotion recognition in South Korea. Ildiko Plájás aims to finish a visual ethnography focused on a Romanian computer vision lab, which should conclude next year. The PI is researching surveillance tech partnerships in Amsterdam.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

We have advanced beyond the current state of the art by advancing several novel working hypotheses on the uses of computer vision in the field of security, and developing innovative multimodal methods in data visualization, specifically within International Relations (IR) and digital social sciences. Our introduction of time-based diagrams offers a novel perspective on data analysis not previously utilized. Moreover, our integration of both quantitative and qualitative data collection methods represents a new approach in critical scholarship which transcends the traditional division between quantitative and qualitative social science.

Security Vision project logo