Exploration and Visual Reconnaissance of Disaster Sites by Drones – EPERVIER

Drone surveillance in disaster situations has become an essential tool for damage assessment, human life rescue, and investigation of causes. Thanks to their ability to quickly access areas that are difficult or dangerous for humans, drones can collect critical real-time data, such as aerial images, videos, and 3D maps. This information is crucial for coordinating rescue efforts, locating victims, and assessing damaged infrastructure. In this context, the project aims to develop an efficient and operational data processing pipeline, managing the flow from data acquisition to object/person detection and localization, using a multidisciplinary approach based on the latest technologies and artificial intelligence. More specifically, EPERVIER seeks to address the challenges faced by Gendarmerie Nationale, particularly in improving video quality to provide evidence in disaster site exploration and crisis management. Furthermore, the project aims to recommend the capabilities that future drones will need to accomplish these missions under even more challenging conditions, such as unstable bandwidth. This aligns with the research axis H.17 of the AAPG2025 particularly by contributing to person or fragment detection, crisis management, and resilience in extreme situations requiring expert assessment.