08.24.2020 news

DeepCube

Agenium Space develops a service of Deep Learning at the edge

ESA supports Agenium Space in the development of its service of Deep Learning at the edge, aiming at simplifying fitting DNN (Deep Learning Neural Networks) in on board HW, to make better use of AI on space missions.

The objective of the activity is to design DNN simplification strategies to provide services to the future customers.

A DNN simplification software will be developed as a tool to support Agenium Space business strategy. Agenium Space's DeepCube software will simplify DNN models to reduce resources required for inference execution, considering the capabilities of existing HW (COTS and space-qualified HW).

The role of the service is to support data processing engineers in adapting powerful DNN for image analysis to reduced HW resources for on-board processing. This service will be available S2 2021.

Agenium Space is already involved in different studies with ESA and CNES dealing with Deep Learning at the edge:

  • R&T study for CNES regarding algorithm part of Smart Payload using DL,
  • Cortex project with ESA (Phi-Lab, Open Call EOEP-5): to define a workflow easing the integration and reduction of complex Deep Neural Networks (DNN) models on Soc-FPGA platforms.

DeepCube project is funded by ESA GSTP Program, with the contribution of the French delegation (CNES). Agenium Space co-invests in the project.

Feel free to look at their web site to read more about this project or to join Agenium Space’s team. If you want to be kept informed of the latest news do not hesitate to follow the Linkedin account of Agenium Space.

We use cookies to personalize content and to analyze our traffic. Please decide if you are willing to accept cookies from our website.