AI4Platform: INFER – artIficial iNtelligence for Food sEcuRity –

The artIficial iNtelligence for Food sEcuRity – INFER – Project aims to foster adoption of AI models for Earth Observation (EO) applications. Many initiatives have been recently launched by ESA to augment the availability of training data, to support new ideas through creation and management of challenges, and to incrementally adopt standards that facilitate interchange and reuse of resources. This project has to goal to focus on a more operational aspect, e.g. to enhance existing cloud-based processing oriented platforms (e.g. TEPs, DIAS and other initiatives) by adding AI specific capabilities. These capabilities include the possibility of performing the following:

• Facilitate creation or improvement of existing models, by running on an already established platform providing seamless access to data and scalable processing resources.
• Facilitate access to existing training datasets and exchange of training datasets within a community.
• Provisioning of AI-oriented tools in a unique environment which serves as an experimental lab for Data Scientists.
• Supporting the possibility for third parties to deploy and make available AI models and training datasets with a predefined remuneration model.

We are proposing a simple and robust implementation approach, which is based on the FS-TEP. This platform is already listed in the services offered by the NOR. The platform will be enhanced with additional AI-related features and with an enhanced module supporting fine grained accounting to enable the required business models.

The consortium will be led by CGI, who has extensive expertise in developing and operating distributed EO-processing platforms and will integrate a set of tools developed during previous activities, including an ESA activity sponsored by the Open Call mechanism, during which a suite of AI-oriented tools were developed and demonstrated. The consortium will be complemented by 2 partners with solid expertise in EO and with a focus on the AI adoption, namely:

•  KP Labs, very active in the field of AI activity in ESA will implement a showcase based on super-resolution for hyperspectral data, by leveraging the finding of other related projects. KP Labs will be the independent entity validating the service
• Vista, the prime contractor of the FS-TEP will bring another showcase focusing on crop mapping and will also complement the consortium with the relevant business-oriented expertise derived from selling commercial EO services to third parties and from the direct management of the NOR offering