At dAIEDGE, our Meet the Winners series continues with a project that pushes AI beyond terrestrial boundaries and into orbit. In this edition, we feature Konstantinos Pikounis, winner of the Second Open Call, and his project AMFITRITE.
AMFITRITE sits at the intersection of Earth observation and artificial intelligence, addressing a critical environmental challenge: the detection of harmful algal blooms (HABs), which pose serious risks to aquatic ecosystems and public health. The project aims to enable fast, automated detection of these events by leveraging AI directly on satellite systems.
To achieve this, AMFITRITE focuses on building two specialized annotated satellite image datasets, one dedicated to water bodies and another for open-water environments. Based on these datasets, the project develops lightweight deep learning models, specifically CNN architectures designed to detect algal blooms efficiently under strict computational constraints.
A key innovation of the project is its on-board satellite processing approach. Instead of transmitting large volumes of raw imagery to Earth, AMFITRITE enables data to be processed directly in orbit. Only critical alerts and relevant information are sent back, significantly reducing bandwidth requirements while enabling faster response times for environmental monitoring.
During his journey within dAIEDGE, Konstantinos developed his project in collaboration with Ubotica, acting as hosting institution. We would like to sincerely thank Ubotica for their support, expertise and guidance throughout the programme, particularly in the domain of satellite-based AI systems.
A central element of this collaboration was the opportunity to move from theoretical design to real deployment considerations. Throughout the seven-month programme, AMFITRITE was progressively validated within the dAIEDGE-VLab environment, allowing Konstantinos to test performance under realistic computational constraints and better understand the trade-offs between model accuracy and on-board efficiency.
This hands-on approach was key in shaping the final direction of the project, reinforcing the importance of lightweight models and real-time execution capabilities for space-based AI applications.
AMFITRITE reflects the mission of dAIEDGE: enabling impactful, deployable AI solutions that address real-world challenges at scale, from Earth to orbit.