Kalkan, Mahmut FurkanKalkan, Ibrahim HalilYilmaz, Necip FazilDispinar, DeryaKahruman, CemYavuz, Abdulcabbar2026-02-082026-02-0820252193-97642193-9772https://doi.org/10.1007/s40192-025-00422-5https://hdl.handle.net/20.500.12885/5590This study showcases a proof-of-concept of an artificial intelligence-driven analytical technique that facilitates the automated extraction of significant quantitative data from microstructural images. Semantic segmentation and classification were conducted on eutectic Si particles and dendritic architectures utilizing microscopic images of AlSi9 alloys with varying Sr ratios. Following segmentation, characteristics including area, aspect ratio, maximum Feret diameter, circularity and SDAS were assessed automatically, and the resulting values were compared with both literature and manual measurements. The samples were effectively categorized based on their alteration levels using a CNN-based classification algorithm. This technology provides significant temporal and financial benefits for microstructural investigation by executing the entire procedure autonomously and expeditiously. The minimal error rates and elevated accuracy findings demonstrate the usefulness and dependability of the devised method for automated microstructural analysis. This paper exemplifies the application of artificial intelligence-driven microstructural analysis techniques in materials science, addressing a significant gap in the literature.eninfo:eu-repo/semantics/closedAccessAluminum silicon alloyDeep learningSemantic segmentationCNNMicrostructure analysisArticle10.1007/s40192-025-00422-5144631642WOS:0015737970000012-s2.0-105016741785Q3Q2