Home > Archive > 2020 > Volume 10 Number 1 (Jan. 2020) >
IJMLC 2020 Vol.10(1): 128-133 ISSN: 2010-3700
DOI: 10.18178/ijmlc.2020.10.1.909

A Deep Neural Network for Pixel-Wise Classification of Titanium Microstructure

Sirodom Mongkhonthanaphon and Yachai Limpiyakorn

Abstract—In recent decades, deep learning has been widely used for automatically classifying images with high accuracy. Semantic segmentation is a type of deep learning that focuses on classifying every pixel into classes. In Metallurgy domain, Titanium and its alloys are considered an ideal biomaterial that suit for implant material. In this research, deep learning is thus introduced as a means to automate the pixel-wise classification of Titanium microstructure in order to reduce the time and uncertainties during human inspection for quality control. The method applies a semantic segmentation technique, fully convolutional neural network, implemented with the U-net architecture. The proposed approach has investigated the integration of Fine-tuning to the U-net architecture for improving the model performance. The dataset of Ti-6A1-4V microstructure images is augmented using elastic deformations. The results report slightly increase of accuracy, while the training time is much faster than training from scratch.

Index Terms—Fine-tuning, fully convolutional neural network, semantic segmentation, titanium microstructure.

The authors are with the Department of Computer Engineering, Chulalongkorn University, Bangkok 10330, Thailand (e-mail: seerowdom@gmail.com, yachai.l @chula.ac.th).

[PDF]

Cite: Sirodom Mongkhonthanaphon and Yachai Limpiyakorn, "A Deep Neural Network for Pixel-Wise Classification of Titanium Microstructure," International Journal of Machine Learning and Computing vol. 10, no. 1, pp. 128-133, 2020.

Copyright © 2020 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).

 

General Information

  • E-ISSN: 2972-368X
  • Abbreviated Title: Int. J. Mach. Learn.
  • Frequency: Quaterly
  • DOI: 10.18178/IJML
  • Editor-in-Chief: Dr. Lin Huang
  • Executive Editor:  Ms. Cherry L. Chen
  • Abstracing/Indexing: Inspec (IET), Google Scholar, Crossref, ProQuest, Electronic Journals LibraryCNKI.
  • E-mail: ijml@ejournal.net


Article Metrics in Dimensions