MetaScope: A Model for Microstructural Analysis of Carbon Steels Using Machine Learning Techniques
Resumo
Microstructural characterization in metallography often relies on manual analysis, which introduces bias and reduces reproducibility. To overcome this, we propose MetaScope, a deep learning model for pixel-level segmentation of ferrite and pearlite in SAE 1020 and 1045 steels. Using 2,200 images with masks generated by a Gaussian Mixture Model, the model combines an U-Net architecture with an EfficientNetB0 encoder. This approach demonstrates the potential of using machine learning techniques to support metallographic workflows and future expansion to other materials.Referências
Amano, N., Lei, B., Müller, M., Mücklich, F., and Holm, E.A. (2025) “Microscopy modality transfer of steel microstructures: Inferring scanning electron micrographs from optical microscopy using generative AI”, Materials Characterization, Elsevier, Netherlands. DOI: 10.1016/j.matchar.2024.114600.
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., and Fei-Fei, L. (2009) “ImageNet: A large-scale hierarchical image database”, IEEE Conference on Computer Vision and Pattern Recognition, IEEE, USA. DOI: 10.1109/CVPR.2009.5206848.
Gargiulo, F., Fontaine, S., Dubois, M., and Tubaro, P. (2022) “A meso-scale cartography of the AI ecosystem”, arXiv preprint, arXiv:2212.12263, Cornell University, USA.
Ishtiaq, M., Inam, A., Tiwari, S., and Seol, J.B. (2022) “Microstructural, mechanical, and electrochemical analysis of carbon doped AISI carbon steels”, Applied Microscopy, 52:10. SpringerOpen, Republic of Korea. DOI: 10.1186/s42649-022-00079-w.
Motyl, M., Madej, Ł. (2022) “Supervised pearlitic–ferritic steel microstructure segmentation by U-Net convolutional neural network”, Archives of Civil and Mechanical Engineering. DOI: 10.1007/s43452-022-00531-4.
Luengo, J., Moreno, R., Sevillano, I., Charte, D., Peláez-Vegas, A., Fernández-Moreno, M., Mesejo, P., Herrera, F. (2022) “A tutorial on the segmentation of metallographic images: Taxonomy, new MetalDAM dataset, deep learning-based ensemble model, experimental analysis and challenges”, Information Fusion. DOI: 10.1016/j.inffus.2021.09.018.
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., and Fei-Fei, L. (2009) “ImageNet: A large-scale hierarchical image database”, IEEE Conference on Computer Vision and Pattern Recognition, IEEE, USA. DOI: 10.1109/CVPR.2009.5206848.
Gargiulo, F., Fontaine, S., Dubois, M., and Tubaro, P. (2022) “A meso-scale cartography of the AI ecosystem”, arXiv preprint, arXiv:2212.12263, Cornell University, USA.
Ishtiaq, M., Inam, A., Tiwari, S., and Seol, J.B. (2022) “Microstructural, mechanical, and electrochemical analysis of carbon doped AISI carbon steels”, Applied Microscopy, 52:10. SpringerOpen, Republic of Korea. DOI: 10.1186/s42649-022-00079-w.
Motyl, M., Madej, Ł. (2022) “Supervised pearlitic–ferritic steel microstructure segmentation by U-Net convolutional neural network”, Archives of Civil and Mechanical Engineering. DOI: 10.1007/s43452-022-00531-4.
Luengo, J., Moreno, R., Sevillano, I., Charte, D., Peláez-Vegas, A., Fernández-Moreno, M., Mesejo, P., Herrera, F. (2022) “A tutorial on the segmentation of metallographic images: Taxonomy, new MetalDAM dataset, deep learning-based ensemble model, experimental analysis and challenges”, Information Fusion. DOI: 10.1016/j.inffus.2021.09.018.
Publicado
12/11/2025
Como Citar
BELLINI, Lucian Fernando; SANTOS, Éric Cassiano Drey dos; KANG, Yeun Haur; HAUPT, William; HÖLBIG, Carlos Amaral.
MetaScope: A Model for Microstructural Analysis of Carbon Steels Using Machine Learning Techniques. In: ESCOLA REGIONAL DE APRENDIZADO DE MÁQUINA E INTELIGÊNCIA ARTIFICIAL DA REGIÃO SUL (ERAMIA-RS), 1. , 2025, Porto Alegre/RS.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2025
.
p. 256-259.
DOI: https://doi.org/10.5753/eramiars.2025.16633.
