Description du projet
Les technologies d’intelligence artificielle de pointe peuvent-elles stimuler le marché de l’observation de la Terre?
Les technologies d’observation de la Terre (OT) prennent du galon avec le projet SnapEarth, financé par l’UE. Ce projet permettra au marché de passer de l’analyse des mégadonnées OT à l’analyse des données en mouvement: le fast data. Le projet cherche à utiliser les données de manière efficace pour prendre des décisions importantes et anticiper les futures priorités en influençant l’intelligence artificielle, les technologies et les outils du cloud. La grande nouveauté vient de EarthSignature, un service d’analyse de données novateur qui a pour objectif d’extraire automatiquement des informations sémantiques à partir d’images satellites. Les informations extraites sont ensuite indexées par le moteur de recherche Qwant, la base de données étant enrichie en temps réel et rendue facilement accessible à un large éventail de communautés d’utilisateurs.
Objectif
SnapEarth will unlock new value, derive actionable service ideas on top of EO big data collections, and anticipate future
priorities by leveraging cutting-edge Artificial Intelligence and Cloud technologies and tools. Thanks to an innovative cloud
agnostic product, SafeScale that is already operational on Copernicus RUS project led by CSSI, users and service providers
on top of SnapEarth will benefit, in a transparent way, from processing platforms and data collections provided by any of the
future C-DIAS and any cloud provider. This cloud brokering solution is providing a performant, cost effective environment,
also protecting their investment, for the future third parties which are building their own services. The major breakthrough of
SnapEarth comes with a new data analytics service, EarthSignature, which aims to automatically extract semantic
information from satellite imagery. The extracted semantic information will be indexed by QWANT search engine and then
be easily accessible to a wide range of user communities. EO experts and third parties will be able to train deep learning
processing chains using their database of labelled EO images. The database will be near real time enriched. Therefore,
SnapEarth allows the market move from analysing EO Big Data towards realizing Fast Data. It makes possible to buy basic
imagery analysis as a commodity – much like we buy foundation data today. Several user communities are ready to engage
in this new approach. SnapEarth proposes already several pilots projects. The first one (EarthSearch) will boost QWANT
number of users through access to this wealth of data through natural language. The second one (EarthPress) involves
press users who are very interested having contextual data linked to news. The following ones are linked to several EO
vertical markets: EarthClimate, Agri, and Food. The last one EarthSelf-service is dedicated to professional third parties in the
same model of the DIAS but ensuring independence.
Champ scientifique
- natural sciencescomputer and information sciencesdatabases
- engineering and technologymechanical engineeringvehicle engineeringaerospace engineeringsatellite technology
- natural sciencescomputer and information sciencesdata sciencebig data
- natural sciencescomputer and information sciencesartificial intelligencemachine learningdeep learning
Mots‑clés
Programme(s)
Régime de financement
IA - Innovation actionCoordinateur
57001 Thermi Thessaloniki
Grèce