Presentation Information
[9p-N405-6]Filament 3D reconstruction of MgB2 multi-filament wire using deep learning
〇(BC)Daito Fukui1, Kosuke Ando1, Takahiro Hosokawa1, Mio Ito1, Kotomi Ogi1, Motomune Kodama2, Hideki Tanaka2, Yusuke Shimada3, Akiyasu Yamamoto1 (1.Tokyo Univ. Agricul. Technol., 2.Hitachi,Ltd, 3.Kyushu Univ.)
Keywords:
superconductivity,MgB2,deep learning
In this study, for quantitative analysis of filament microstructure of MgB2 multi filament wire, we first attempted segmentation and 3D reconstruction of filaments by applying a deep learning model to 3D microstructure images obtained by FIB-SEM. The filaments were observed to be generally continuous in the longitudinal direction, suggesting that the segmentation accuracy and continuity of the filaments themselves are good.