At the cutting edge of artificial intelligence and materials science, AID4GREENEST researcher Dr. Daria Tomecka (ePotentia) shared key insights at ML4Spec 2025, the Belgian-German WE-Heraeus Seminar on Machine Learning for Spectroscopy, held from 26–28 May 2025 at Vrije Universiteit Brussel (VUB).
In a contributed talk titled “Real, Fake, or Just Useful? Generation and Evaluation of Synthetic Steel Micrographs with Physical Constraints”, Dr. Tomecka presented recent progress on generating synthetic microstructures of steel.
The research goes beyond visual realism, asking a more fundamental question: are these synthetic images scientifically useful? The presentation focused on developing meaningful evaluation methods for assessing the utility of AI-generated micrographs, an important step for their application in industrial materials development.
In addition to the talk, Dr. Tomecka also presented a poster titled “Generating, Sorting and Standardising Microscopy Data with Artificial Intelligence.” This work, conducted in collaboration with Ghent University (UGent) and OCAS, showcases how AI tools can help streamline and harmonise microscopy data, crucial for accelerating model-based materials design across the steel sector.
Both contributions were part of the AID4GREENEST project, a Horizon Europe initiative dedicated to using AI to develop rapid characterisation methods and modelling tools for steel. The project’s goal is to support more sustainable, efficient and innovative steel production processes across Europe.
For more information about the event, visit: ml4spec25.sciencesconf.org