LiDAR technologies, photogrammetry, and 3D scanners now make it possible to produce enormous volumes of data describing our cities, coastlines, infrastructure, and cultural heritage. However, automatically exploiting and analyzing these data remains a major challenge. The ModAM³ project aims to develop a new generation of artificial intelligence methods for analyzing 3D data and automatically detecting objects and changes in the environments under study. Within this project, GREYC (UMR CNRS 6072) proposes to introduce new multi-scale graph-based representations adapted to massive and temporal data. Data processing tasks, such as localization, classification, and object recognition, will be addressed through the development of novel fundamental methods for semi-supervised and unsupervised learning. These methods will be based on new classes of partial differential equations on graphs and new graph neural network models based on diffusion processes. The developed algorithms will be applied to several application domains, particularly cultural heritage, with a focus on Pointe du Hoc, a major historical site of the Second World War.GREYC already has expertise in 3D reconstruction and 3D printing for cultural heritage applications, with the aim of facilitating access to and work by researchers and conservators, as well as improving accessibility for visually impaired people.

Plus d’informations :
[Website EURAXESS]