Crowded gatherings in public spaces are common in Europe, whether they are festive events (festival of lights in Lyon, Hafengeburstag in Hamburg, European football cup, etc.), or demonstrations ( political, trade-union, etc.). S2UCRE's goal is to help make them safer.
Large urban gatherings are a challenge for the teams in charge of safety and security. <br />Large and crowded urban gatherings in public spaces are common in Europe, whether they are festive events (sport, music, etc.) or demonstrations (political, trade union, etc.). <br />These gatherings are a challenge for the authorities because they must ensure that <br />• The emergency services (fire brigade, red cross, etc.) can take care of the physical safety of the attendees (avoiding accidents in stampede, health risks, etc.), <br />• The police can maintain public order, and in particular that they can follow and find individuals who have been caught committing misdeeds (for example having caused violent clashes), individuals known to be dangerous and likely to commit an attack, and also the people lost in the crowd that their relatives are looking for. <br />As large urban gatherings are most often festive events, it is also desirable that security does not unnecessarily disturb the peace and enjoyment of the participants in the event, and that it respects the spirit of the gathering. <br />The objective of the S2UCRE project, supported by the ANR and the BMBF, is to help maintain security and safety in large urban crowded events.
To improve the safety of large gatherings, S2UCRE relies on:
• A command centre
• Field staff and their communication equipment
o Law enforcement
o Emergency personnel
o Possibly the organizers' security team
• A video protection network.
S2UCRE is developing innovative methods for real-time analysis of available videos:
• Measuring the density and direction of the crowd, forecasting its evolution, and piloting actions to avoid saturation situations.
• Follow-up of people who have been identified by the police as a significant security concern, or search for people for security reasons.
These analyses are supplemented by field observations: staff can report information, and verify, correct, or add to the information coming from the video analysis.
S2UCRE improves cooperation between the various security players, by allowing them to interact using multimedia tools under the control of the command centre. S2UCRE is also extending communication to citizens, whether or not participating in the event, to warn them in case of danger through SMS.
S2UCRE allows precise geolocation of security personnel and events, in order to optimize interventions.
Finally S2UCRE clarifies the legal context of the technologies it implements, and studies the sociological acceptability of these technologies.
Results obtained by the German partners of S2UCRE.
S2UCRE has met its target in terms of crowd video analysis: technologies for measuring crowd density and movements from cameras have been developed and optimized, as well as drone-based techniques for the prior calibration of the cameras overlooking the place of the gathering. Prediction models were developed to better guide the crowd in the event of risk: it appeared that these models were very complex, and could also be useful in helping to respect physical distancing. In addition, communications with emergency personnel have been secured to ensure their efficiency, and also to comply with personal data protection requirements. People in charge of crowd surveillance were interviewed, and they judged that the S2UCRE toolkit would have an overall favourable impact on their work.
Results obtained by the French partners of S2UCRE.
The dialogue infrastructure developed in S2UCRE allows security actors (security and safety teams) to have at all levels an up-to-date and comprehensive overview of the situation around them. The actors in the field themselves contribute to improving this overview, and their missions are optimized thanks to geolocation. Citizens are also taken into account thanks to alerts to the population and the connection to the European emergency telephone number 112.
The video analytics developed in S2UCRE make it possible to find people by their description and their face, and to follow them from one camera to another thanks to their face or body image, even in a crowd situation. The legal study clearly shows that their use is restricted by the principle of proportionality, that is to say that the implementation of these technologies must be restricted to critical cases, and controlled by competent entities. Legal regulations and ethical principles had a huge impact in S2UCRE: for instance, the tests carried out were only on videos where all the participants were voluntary and therefore cooperative.
The perspectives of the S2UCRE project are, with regard to the communication aspects, to exploit the new means developed for rescue and police services, corresponding to S2UCRE use cases.
Regarding video surveillance aspects, the S2UCRE project outcome, corresponding to the S2UCRE use cases, is usable depending on legal and acceptability aspects. Indeed, technologies can certainly still be improved, but this only makes sense if they are acceptable from the legal and societal perspective. As of today (Q1 2021), the technologies developed in S2UCRE can be used during an investigation and therefore after the event, under the control of the justice system, and of course without any automatic decision being taken (any decision must be validated by a human being). It is the exploitation which is considered for the moment.
The publications that have taken place thanks to S2UCRE (ANR side) are mainly
• of a legal nature, in particular with regard to the study of the use of drones for the maintenance of public order, a use which was excluded in S2UCRE,
• of a technical nature, relating to
o the calibration of a mobile camera using a fixed camera (when the images they provide partially show the same place),
o the geolocation of an element of a camera image, when it is calibrated with possible uncertainties
In both cases, techniques dedicated to information fusion are proposed.
Mass gatherings at large events and demonstrations in public urban spaces are an expres-sion of freedom and openness in European societies. German and French citizens enjoy sport events, like marathon runs, festivals like the Cannstatter Wasn in Stuttgart, the Munich Oktoberfest, or Francofolies de la Rochelle, Feria de Bayonne, Fetes des Lumières in Lyon and many more. Festivals and politically motivated demonstrations alike involve threats for the safety and security of citizens. Therefore, organizers and public authorities are facing new challenges and are in need of new solutions. In this project, partners from academia, civil security forces, one large security industry partner and several small enter-prises join forces to investigate and evaluate solution strategies, methods and concepts.
S²UCRE stands for Safety & Security of Urban Crowded Environments. The project addresses mass gatherings in complex scenarios and widespread urban environments. In such situations, the crowd is often inhomogeneous in terms of density and motion dynamics. In addition, pedestrians are distributed over large areas. The surveillance of the course of the event is challenging for Law Enforcement Agencies (LEAs) even without disturb-ances. An up-to-date overview is often missing. Fast reaction in case of accidents, offen-sive behaviour, or crowd densities that trap people and endanger lives (often referred to as “panic situations”) is difficult, pre-emptive measures seem impossible.
To overcome these problems, the S²UCRE project aims to interconnect crowd monitoring based on video analytics with simulation-based methods for short-term prediction of crowd behaviour. Two concrete events are selected as scenarios: the Hamburg Harbour Festival (1.5 million visitors) and the May 1st demonstrations in Paris (12000 attendees). S²UCRE furthermore intends to bridge the gap between the macro level of crowd moni-toring and observation of individuals or groups when there is reasonable ground for suspicion. Carrying out research on five interdependent support technologies for LEAs, event organizers, security staff and rescue teams, S²UCRE will improve the collaboration of stakeholders responsible for security. Studies on human factors form an integral part of the technical work packages and thus go beyond accompanying re-search. A fundamental work package is dedicated to legal and ethical aspects. The S²UCRE demonstrator will integrate the support technologies, thus providing a proof of concept for predictive analytics tools. In particular, ways to guide crowds based on predic-tive simulation will be explored. The five technology areas are:
• Video-based crowd monitoring of distributed urban environments: crowd densities and crowd dynamics.
• Short-term prediction of crowd behaviour for fast and efficient evacuation.
• Semi-automated suspicious behaviour analysis for security applications.
• Detection & (geo-) localisation of perpetrators.
• Self-localisation of security and rescue team members and a communication platform for geo-registered information exchange between security and rescue personnel.
Unlike other projects, S²UCRE will not develop these technologies individually but will focus the research on how to integrate a continuous workflow into a single system. This will combine the current security and safety situation with short-term simulations of the crowd that are used to prevent critical conditions.
S²UCRE envisions an integrated surveillance and prediction system that addresses safety and security and that is tested on concrete scenarios.
Madame Claude BAUZOU (IDEMIA IDENTITY & SECURITY)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
IPSS INGENIERIE POUR SIGNAUX ET SYSTEMES
UPSud/SATIE Université Paris Sud/Laboratoire des Systèmes et Applications des Technologies de l'Information et de l'Energie
MPH IDEMIA IDENTITY & SECURITY
CERAPS Centre d'Études et de Recherches Administratives, Politiques et Sociales
PPP Prefecture de Police de Paris
Help of the ANR 1,042,506 euros
Beginning and duration of the scientific project: May 2017 - 36 Months