Atos’ solution simplifies and accelerates the handling of user reports

Atos has announced its selection by Métropole Aix-Marseille-Provence for the implementation of an industrialised solution for contextual image analysis and its integration with various metropolitan devices as part of the FIDAMIA project. These devices allow users to report equipment malfunctions and initiate the resolution process. The contract encompasses the development, deployment and maintenance of a self-learning artificial intelligence (AI) model, as well as training for Métropole agents.

A New Digital Service Integrated into the ‘Ma Métropole Dans Ma Poche’ Application

The contextual image analysis solution deployed by Atos streamlines the citizen alert system, expediting interventions by the technical services of Métropole Aix-Marseille-Provence and its subcontractors. Today, citizens can choose from three automatically detected categories to provide their reports, whereas previously they had to select from over 30 categories. The analysis, qualification and allocation of intervention requests to the appropriate domains of responsibility are now facilitated. This results in reduced processing times through OSIS (the tracking and instruction tool for reports) and shorter intervention times for the technical services.

The diagnostic assistance based on Atos’ artificial intelligence platform enables the identification or confirmation of the type of report and precise qualification of the intervention request and malfunction, whether it pertains to cleanliness, roads, signage, or illicit displays, among others.

Accelerated Response Times

Atos’ solution ensures a response time per request of less than one second and a 90% accuracy rate for assigning citizen requests to the correct category. These performance indicators surpass the initial specifications.

To prepare the training data for the AI model, Atos had access to a dataset of approximately 360,000 photos and associated typology as well as a database of over 2 million processed reports representing 1.5 million images.

An Industrialised Approach

In order to meet Métropole Aix-Marseille-Provence’s objective of providing a higher quality service to users, Atos proposed a tool-driven approach to industrialise AI models. This approach relies on Machine Learning Operations (MLOps), a set of practices aimed at automating, deploying, and monitoring AI models in a production environment.

In line with digital sobriety, Atos recommended a lightweight approach based on two virtual machines, while ensuring the industrialisation and rapid deployment of AI models.

“Recently elected the European Capital of Innovation, Métropole Aix-Marseille-Provence is committed to offering residents a high-performance user experience available at all times. The combination of artificial intelligence and Atos’ expertise will allow users to simplify their interactions with our technical services, improve intervention times, and contribute to enhancing the quality of life for all,” explains Arnaud Mercier, Councillor of the Métropole, representative for the Digital Métropole, Public Data Policy, Innovation, and User Experience.

“This project undertaken for Métropole Aix-Marseille-Provence showcases our ability to develop and implement AI models that meet the expectations of local authorities. Leveraging our expertise in AI learning solutions, intelligent data platforms and virtual infrastructures, we delivered an operational tool in less than 6 months,” says Laurent Laffere, Director of the Public Sector, Defence and Health at Atos.

Thanks to the FIDAMIA project, Métropole Aix-Marseille-Provence has been named the “Territoires Numériques” winner by the Interministerial Directorate of Digital Technology. This distinction recognises digital projects subsidised through the France Relance program.

Top image: AMP Métropole