Testing automated video analysis to capture discrimination in Germany

Data-Method-Monitoring Cluster

Project head: Dr. Jörg DollmannDr. Jannes Jacobsen

Associates: Dr. Julia BehrmanDr. Doron Shiffer-Sebba

Running time January 2024 until December 2024
Status Completed project

This project aims to explore whether and under what conditions automated video analysis can be utilized for discrimination detection in Germany. A so-called "Proof of Concept" will be developed within this project. This means that based on various test situations, an experimental setup will be validated, and subsequent legal questions regarding data protection and the protection of personal rights will be discussed. Therefore, at the project's conclusion, there will be a guideline for the application and processing of video material from public spaces for social science in Germany.

As a foundation and test case, an experimental arrangement from the USA will be replicated in Germany. In the existing approach, so-called "Racial Avoidance" was measured using automated procedures. "Racial Avoidance" was operationalized based on the distance individuals maintain from racialized persons on the street.

To implement this method, public spaces will be filmed, and pairs of individuals, belonging either to a racialized group or a non-racialized group, will create artificial situations of spatial restriction. The video (in combination with computer vision technology) then enables the measurement of the distance between individuals. The appearance of the two individuals being studied can be varied and combined in various experimental setups (e.g., gender, national or religious symbols, physical impairment).

Funding: Federal Ministry for Education, Family Affairs, Senior Citizens, Women and Youth (Institutional funding)