TY - JOUR
T1 - Analysis of TEM images of metallic nanoparticles using convolutional neural networks and transfer learning
AU - Koyama, Akira
AU - Miyauchi, Shoko
AU - Morooka, Ken'ichi
AU - Hojo, Hajime
AU - Einaga, Hisahiro
AU - Murakami, Yasukazu
N1 - Funding Information:
The authors are grateful to Mr. J. Ohta. Ms. A. Sato, and Drs. T. Yamamoto, T. Tamaoka and R. Aso for their helpful comments on the image analysis, and to Drs. Y. Cho and Y. Nakajima for their assistance in acquiring the TEM images. This work was supported in part by JST CREST (grant number JPMJCR1664) and JSPS KAKENHI (grant number 18H03845).
Publisher Copyright:
© 2021 Elsevier B.V.
PY - 2021/11/15
Y1 - 2021/11/15
N2 - Convolutional neural networks (CNNs) pretrained by transfer learning were applied to the analysis of transmission electron microscopy (TEM) images of nanoparticles. Specifically, TEM images of non-magnetic Pt nanoparticles dispersed on a thin TiO2 crystal foil were classified using CNNs. Although the number of learning data (50≤ N≤350) was several orders of magnitude smaller than the quantities normally employed in conventional CNN analyses, the present CNN model was able to carry out image classification with 94% accuracy (average of 25 results) after the convolutional layers were pretrained by transfer learning and fine tuning. This method represents a promising tool for TEM studies of both non-magnetic and magnetic nanoparticles which make emergence of rich material functions.
AB - Convolutional neural networks (CNNs) pretrained by transfer learning were applied to the analysis of transmission electron microscopy (TEM) images of nanoparticles. Specifically, TEM images of non-magnetic Pt nanoparticles dispersed on a thin TiO2 crystal foil were classified using CNNs. Although the number of learning data (50≤ N≤350) was several orders of magnitude smaller than the quantities normally employed in conventional CNN analyses, the present CNN model was able to carry out image classification with 94% accuracy (average of 25 results) after the convolutional layers were pretrained by transfer learning and fine tuning. This method represents a promising tool for TEM studies of both non-magnetic and magnetic nanoparticles which make emergence of rich material functions.
UR - http://www.scopus.com/inward/record.url?scp=85109431464&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85109431464&partnerID=8YFLogxK
U2 - 10.1016/j.jmmm.2021.168225
DO - 10.1016/j.jmmm.2021.168225
M3 - Article
AN - SCOPUS:85109431464
SN - 0304-8853
VL - 538
JO - Journal of Magnetism and Magnetic Materials
JF - Journal of Magnetism and Magnetic Materials
M1 - 168225
ER -