Andrea Cavallini, AI/ML Competence Area Lead and VIPER Project Manager at Starion, shares news of the launch of the VIPER Project, a ground-breaking artificial intelligence (AI) initiative funded by the Italian Space Agency (ASI) and developed by Starion Italia in collaboration with Nautilus – Navigation in Space Srl.
Why VIPER is revolutionary
Many AI solutions today address individual tasks in isolation – such as denoising, super-resolution or object detection – without considering the inefficiencies and missed opportunities of this fragmented approach. Each task is often developed and optimised separately, leading to siloed workflows that fail to unlock the true potential of AI systems.
VIPER – or, to give it its full name, the Versatile Intelligent Processing Enhancement and Refinement chain – changes the game by introducing a holistic, end-to-end methodology. Instead of treating these tasks as standalone challenges, VIPER integrates them into a unified, modular chain. This transformative approach allows us to seamlessly combine data input, enhancement, enrichment and analysis into a single, cohesive process. The result? Maximised AI exploitation at every stage, streamlined workflows and a significant leap in efficiency and adaptability across applications.
A ground-breaking step forward in AI development
The “build once, use many” philosophy at the heart of VIPER delivers a reusable and adaptable AI processing chain. This approach allows for greater scalability and ensures that AI capabilities can be applied across a wide range of computer vision tasks without having to rebuild systems from scratch. Whether the challenge is Earth observation (#EO), medical imaging or any other domain, VIPER empowers users with:
- Modular flexibility: Activate or deactivate modules to fit specific needs.
- Cross-platform functionality: Optimised for both traditional hardware and on-board systems for faster, localised and real-time processing.
- Enhanced efficiency: Tackle entire workflows, reducing cost, time and computational effort.
Proving AI’s power with two diverse use cases
To showcase the versatility and power of VIPER, we have carefully selected two diverse and contrasting use cases:
- Illegal landfill detection: Harnessing satellite EO data to identify and monitor unauthorised waste disposal sites, addressing critical environmental and regulatory challenges.
- Near-Earth object (NEO) tracking: Utilising telescope data to detect and track potentially hazardous celestial objects, contributing to planetary defence and space safety.
These contrasting scenarios highlight VIPER’s adaptability across vastly different domains, demonstrating its ability to tackle unique challenges while maintaining efficiency and effectiveness.
Progress so far
We kicked off the project on 18 December 2024, starting with on-ground activities. While Phase 1 focuses on demonstrating the chain’s capabilities, we are already preparing for on-board applications in Phase 2, showcasing how AI can deliver real-time, actionable insights in space.
A special thanks to Nautilus – Navigation in Space Srl, our esteemed partner, contributing their expertise in synthetic dataset generation and hardware-in-the-loop test environments, critical to testing and optimising VIPER’s embedded AI systems.
AI exploitation at its core
VIPER is not just a tool; it’s a leap forward in exploiting AI to its fullest potential. By rethinking how we approach image processing, we’re setting new standards for efficiency, scalability and adaptability in AI applications. From reducing the time and cost of development to opening new possibilities in AI-powered analysis, VIPER exemplifies what’s possible when we fully harness the capabilities of artificial intelligence.
Follow us on LinkedIn to stay updated on the progress of the VIPER project. And feel free to contact me for any insights, collaborations or discussions on how this ground-breaking approach to AI exploitation can transform complex workflows across industries!