Victor Maghin: The blind spots of the smart camera. Scientific filters, a confined debate and a discreet market

Surveillance cameras have been a feature of our urban – and even rural – landscapes for several years now. However, these devices have not resolved all the issues that led to their deployment in many public spaces. One of the solutions proposed to address this shortcoming of video surveillance is therefore to add automatic detection software to the existing system. These additional modules come in various forms: people counting, vehicle type detection, fly-tipping detection, analysis of abnormal situations, facial recognition, etc. The first applications of these new forms of video surveillance took place in France, although they are not as widespread as in other countries.

This thesis aims to offer initial reflections on these new surveillance tools and the context in which they operate, examining them from three main perspectives. The first of these concerns the scientific developments that contribute to the development of these new surveillance solutions. The second focuses on the public debate surrounding automated video surveillance and the specific form that this debate takes. The third and final aspect examines the uses, mostly experimental, taking place in France, as well as the objectives and conditions of these applications.

With regard to the first aspect, this study aims to demonstrate the particular importance of infrastructure to the success of these artificial intelligence tools. Whether at the development stage or during use, artificial intelligence relies heavily on all the tools and resources deployed in its operation. Data, networks and sensors act as filters through which the software’s processing passes, imposing technical limitations. Regarding the second aspect, the research has explored the issue of how these technical solutions are presented, promoted or criticised as a problem, both in public debates and within economic spheres. It emerges that, despite widespread criticism within society, economic arguments dominate in official reports. Regarding the third aspect, whilst there is no consensus on the subject of automated surveillance, the thesis analyses the range of strategies and resources deployed to build a market for automated video surveillance nonetheless.

Jury members:
Gilles JEANNOT, ENPC, Thesis Supervisor
; Anne-Cécile DOUILLET, University of Lille, Examiner;
Sylvain PARASIE, Sciences Po Paris, Examiner;
Valérie NOVEMBER, CNRS, Examiner;
Clément MABI, UTC, Examiner