KEYNOTE 2
Web-based AI for real-time prediction of grain size distribution
AI applications are increasingly finding their way into geotechnical engineering The keynote presentation introduces a web-based solution for real-time prediction of the grain size distribution of mineral soils from standardized smartphone images. The prediction accuracy is around 90%. The system is device-independent. The model is available on the Hugging Face platform. The version offers an upgrade of the backbone model, an improved training pipeline, automatic device detection with pixel density estimation, and a PDF report.
Priv.-Doz. Dr. Enrico Soranzo
Enrico Soranzo, Senior Scientist at the University of Sustainable Resources and Life Sciences in Vienna, has been working on the application of machine learning in geotechnical engineering for several years. After completing his doctorate in environmental engineering and his habilitation in geotechnical engineering, he has advanced several research projects on the automation of geotechnical processes, including the use of AI for infrastructure planning and monitoring.
In addition to his scientific work, he brings extensive industry experience from major tunnel construction projects, including the Koralm Tunnel. In his role as deputy director of the institute, he combines theory and practice to develop innovative solutions for sustainable and resilient structures.