In collaboration with MI4People and Alexander Thamm GmbH [at], the AI Competence Centre at Bechtle Bremen, developed a platform at the heart of which is a powerful AI-based tool for identifying marine litter. Leveraging cutting edge computer vision methods and satellite images, it is possible to identify virtually in real time where and to what extent plastic and other litter is polluting our oceans.

A powerful engine room – The Bechtle platform.

The foundations were laid at the Bechtle lab in Bremen where powerful hardware including a Dell XE8545 server with NVIDIA A100 GPUs and a Kubernetes cluster were married together to run the machine learning AI model developed by AT. The software processes the satellite images, which are then analysed by AI before being readied for research institutes, environmental protection organisations and authorities. The platform displays an interactive map showing where in the world marine litter is and can be used to help coordinate clean-up operations.

Our data and AI partner.

While Bechtle provided the AI-compatible infrastructure, AI-specialist AT developed, modified and trained the large language model (LLM) to meet the specific requirements of the project. Working together, the experts have enabled the seamless integration and processing of large volumes of data.


AI has given us the transparency we need to tackle a global problem and is helping us save our oceans.

Torsten Welke, Head of AI Competence Centre, Bechtle IT System House Bremen


The four steps to an AI solution.

The platform was built following a structured, multi-level plan. In step one, the Bechtle Bremen team created a proof of concept with data accessible to the public, after which the model was continually optimised with additional data before the online platform was set up in step three, making the results available to all interested parties.

A use case that extends beyond technology.

This project has underscored how Bechtle uses its AI expertise not only for enterprise digital transformation and automation, but also to help develop solutions that have a far-reaching ecological and social impact.

Discover more about the project and the satellite map here:

MI4People