Ter(r)ra: Risorse, Rischi, Rispetto
Padova, 15-17 settembre 2026
Sponsor Abstracts
How Artificial Intelligence is Changing Geoscience Microscopy
Davide Garoldini1
1Carl Zeiss SpA, Research Microscopy Solutions
Davide Garoldini1
1Carl Zeiss SpA, Research Microscopy Solutions
Keywords: Artificial Intelligence, automated mineralogy, quantitative microscopy.
Artificial intelligence (AI) has rapidly emerged as a transformative force across the geosciences, reshaping the way Earth systems are observed, analysed, and interpreted. The rapid growth of high-resolution datasets, combined with advances in machine learning and deep learning, has enabled researchers to move beyond traditional physics-based approaches and uncover complex, multiscale patterns within geological, atmospheric, and environmental processes. (Zhao et al., 2024).
Alongside this paradigm shift, developments in advanced microscopy and automated analytical workflows are redefining how geoscientific data are generated and interpreted. Thanks to this approach, automated mineralogy - traditionally based on SEM-EDS classification - is extended beyond qualitative or semi-quantitative approaches toward fully quantitative, multi-scale characterization of geological materials. This approach allows the extraction of bulk composition, mineral chemistry, and modal abundance directly from thin sections, significantly streamlining workflows that previously required multiple analytical techniques. (Taylor et al., 2024)
With the advent of deep neural network models, automated mineralogy workflows are further enhanced through AI-driven phase classification, enabling more robust, faster, and highly accurate mineral identification directly from complex datasets. At the same time, the integration of deep learning-based reconstruction methods in X-ray microscopy extends these capabilities into the 3D domain, paving the way to fully automated, high-throughout mineral identification and characterization across multiple scales.
These advances will be illustrated to show how the convergence of quantitative microscopy, large datasets, and AI-driven algorithms is enabling a new generation of correlative, data-rich workflows in geosciences.
Taylor R.J. et al. (2024) - A step forward in quantitative automated mineralogy in 2D and 3D. Geostandards and Geoanalytical Research, 48(3), 579-593, A Step Forward in Quantitative Automated Mineralogy in 2D and 3D - Taylor - 2024 - Geostandards and Geoanalytical Research - Wiley Online Library.
Zhao T. et al. (2024) - Artificial intelligence for geoscience: Progress, challenges, and perspectives. The Innovation, 5(5), Artificial intelligence for geoscience: Progress, challenges, and perspectives - ScienceDirect.
Alongside this paradigm shift, developments in advanced microscopy and automated analytical workflows are redefining how geoscientific data are generated and interpreted. Thanks to this approach, automated mineralogy - traditionally based on SEM-EDS classification - is extended beyond qualitative or semi-quantitative approaches toward fully quantitative, multi-scale characterization of geological materials. This approach allows the extraction of bulk composition, mineral chemistry, and modal abundance directly from thin sections, significantly streamlining workflows that previously required multiple analytical techniques. (Taylor et al., 2024)
With the advent of deep neural network models, automated mineralogy workflows are further enhanced through AI-driven phase classification, enabling more robust, faster, and highly accurate mineral identification directly from complex datasets. At the same time, the integration of deep learning-based reconstruction methods in X-ray microscopy extends these capabilities into the 3D domain, paving the way to fully automated, high-throughout mineral identification and characterization across multiple scales.
These advances will be illustrated to show how the convergence of quantitative microscopy, large datasets, and AI-driven algorithms is enabling a new generation of correlative, data-rich workflows in geosciences.
Taylor R.J. et al. (2024) - A step forward in quantitative automated mineralogy in 2D and 3D. Geostandards and Geoanalytical Research, 48(3), 579-593, A Step Forward in Quantitative Automated Mineralogy in 2D and 3D - Taylor - 2024 - Geostandards and Geoanalytical Research - Wiley Online Library.
Zhao T. et al. (2024) - Artificial intelligence for geoscience: Progress, challenges, and perspectives. The Innovation, 5(5), Artificial intelligence for geoscience: Progress, challenges, and perspectives - ScienceDirect.
***
.jpg)
.jpg)
.jpg)