Automated Microstructure Classification with Class-Specific Segmentation for Titanium Based Composite Materials

Notice

This is an unedited manuscript accepted for publication and provided as an Article in Press for early access at the author’s request. The article will undergo copyediting, typesetting, and galley proof review before final publication. Please be aware that errors may be identified during production that could affect the content. All legal disclaimers of the journal apply.

Year : 2026 | Volume : 14 | 03 | Page :
    By

    Pritam Biswas,

  • Sourav Debnath,

  • Avishek Nath,

  • Paramita Sarkar,

  • Soumyajit Roy,

  • Krishna Kumar Jha,

  1. Student, Department of Computer Science & Engineering, Narula Institute of Technology, Kolkata, West Bengal, India
  2. Assistant Professor, Department of Electrical Engineering, Brainware University, KolkatA, West Bengal, India
  3. Assistant Professor, Department of Computer Science & Engineering, Narula Institute of Technology, Kolkata, West Bengal, India
  4. Assistant Professor, Department of Computer Science & Engineering, JIS University, Kolkata, West Bengal, India
  5. Assistant Professor, Department of Mechanical Engineering, Haldia Institute of Technology, Haldia, West Bengal, India
  6. Assistant Professor, Department of Information Technology, Guru Nanak Institute of Technology, Kolkata, West Bengal, India

Abstract

In engineering, characterisation of microstructure is required to determine and forecast behaviour of titanium alloys. Our proposal in this work has been a deep-learning-based framework in the automatic classification and segmentation of Titanium Based Composite Material. The framework then uses EfficientNetB0 backbone, where we have chosen the backbone to scale the performance of classification and the computational efficiency with the assistance of the transfer learning and the compound scaling. In a bid to enhance the interpretability, we use class-specific segmentation once we have done the classification, which means that the segmentation policy can be adjusted to the morphological features of each predicted class. This causes the identification of duplex, lamellar and martensitic microstructures. The network is trained on the basis of the Adam optimiser and categorical cross-entropy loss, and a stratified approach to data-splitting is used to overcome the imbalance between classes. The model achieved a mean classification accuracy of 96.73% ± 0.67%, with a best observed accuracy of 97.55% and good macro F1-score and AUC results were achieved through experimental assessment on the test set, and these results demonstrate good generalisation across all microstructural classes. In the course of the experimentation, we noted that the proposed framework may facilitate the effective and interpretable microstructural analysis, which is why it may be applied to metallurgical research and quality control in industries.

 

Keywords: Microstructure classification, Titanium Based Composite Material, deep learning, EfficientNetB0, class-specific segmentation, Compound Scaling, Transfer Learning.

How to cite this article:
Pritam Biswas, Sourav Debnath, Avishek Nath, Paramita Sarkar, Soumyajit Roy, Krishna Kumar Jha. Automated Microstructure Classification with Class-Specific Segmentation for Titanium Based Composite Materials. Journal of Polymer & Composites. 2026; 14(03):-.
How to cite this URL:
Pritam Biswas, Sourav Debnath, Avishek Nath, Paramita Sarkar, Soumyajit Roy, Krishna Kumar Jha. Automated Microstructure Classification with Class-Specific Segmentation for Titanium Based Composite Materials. Journal of Polymer & Composites. 2026; 14(03):-. Available from: https://journals.stmjournals.com/jopc/article=2026/view=243383


References

[1] G. L¨utjering and J. C. Williams, Titanium. Berlin, Germany: Springer,2007.

[2] Welsch G, Boyer R, Collings EW, editors. Materials properties handbook: titanium alloys. ASM international; 1993 Dec 31.

[3] Mokhles M, Liu G, Shishavan BH, Dandekar TR, Barouni A, Birosca S. New insights on the deformation mechanism of fretting fatigue in Ti Based Composite Material. Acta Materialia. 2025 Jun 1;291:120998.

[4] Cho KY, Cho S, Jung GY, Eom K. Metal shields with crystallographic discrepancies incorporated into integrated architectures for stable lithium metal batteries. Energy & Environmental Science. 2024;17(9):3123-35.

[5] Yang HB, Yue X, Liu ZX, Guan QF, Yu SH. Emerging sustainable structural materials by assembling cellulose nanofibers. Advanced Materials. 2025 Jun;37(22):2413564.

[6] Wang H, Fu T, Du Y, Gao W, Huang K, Liu Z, Chandak P, Liu S, Van Katwyk P, Deac A, Anandkumar A. Scientific discovery in the age of artificial intelligence. Nature. 2023 Aug 3;620(7972):47-60.

[7] Krizhevsky A, Sutskever I, Hinton GE. ImageNet classification with deep convolutional neural networks. Communications of the ACM. 2017 May 24;60(6):84-90.

[8] Chen LQ, Zhao Y. From classical thermodynamics to phase-field method. Progress in Materials Science. 2022 Feb 1;124:100868.

[9] Liu X, Zhang J, Pei Z. Machine learning for high-entropy alloys: Progress, challenges and opportunities. Progress in Materials Science. 2023 Jan 1;131:101018.

[10] Kibrete F, Trzepieciński T, Gebremedhen HS, Woldemichael DE. Artificial intelligence in predicting mechanical properties of composite materials. Journal of Composites Science. 2023 Sep 1;7(9):364.

[11] A. Chowdhury, E. Kautz, B. Yener, and D. Lewis, “Image-drivenmachine learning methods for microstructure recognition,” Comput.Mater. Sci., vol. 123, pp. 176–187, 2016.

[12] S. Baskaran, R. J. Asaro, and E. A. Holm, “Automated classificationand segmentation of Ti–6Al–4V microstructures using deep learning,”Mater. Sci. Eng. A, vol. 765, pp. 138–149, 2019.

[13] Chen RJ, Ding T, Lu MY, Williamson DF, Jaume G, Song AH, Chen B, Zhang A, Shao D, Shaban M, Williams M. Towards a general-purpose foundation model for computational pathology.

 

Nature medicine. 2024 Mar;30(3):850-62.

[14] Terven J, Córdova-Esparza DM, Romero-González JA. A comprehensive review of yolo architectures in computer vision: From yolov1 to yolov8 and yolo-nas. Machine learning and knowledge extraction. 2023 Nov 20;5(4):1680-716.

[15] Tan M, Le Q. Efficientnet: Rethinking model scaling for convolutional neural networks. InInternational conference on machine learning 2019 May 24 (pp. 6105-6114). PMLR.

[16] Ren Z, Wang S, Zhang Y. Weakly supervised machine learning. CAAI Transactions on Intelligence Technology. 2023 Sep;8(3):549-80.

[17] Moses IA, Reinhart WF. Transfer learning for multi-material classification of transition metal dichalcogenides with atomic force microscopy. Machine Learning: Science and Technology. 2024 Dec 1;5(4):045081.

[18] Kim W, Jung S, Moon Y, Mangum SC. Morphological band registration of multispectral cameras for water quality analysis with unmanned aerial vehicle. Remote Sensing. 2020 Jun 24;12(12):2024.

[19] Shen C, Zhao J, Huang M, Wang C, Zhang Y, Xu W, Zheng S. Generation of micrograph-annotation pairs for steel microstructure recognition using the hybrid deep generative model in the case of an extremely small and imbalanced dataset. Materials Characterization. 2024 Nov 1;217:114407.

[20] Wang K, Gupta V, Lee CS, Mao Y, Kilic MN, Li Y, Huang Z, Liao WK, Choudhary A, Agrawal A. XElemNet: towards explainable AI for deep neural networks in materials science. Scientific reports. 2024 Oct 24;14(1):25178.

[21] Che L, He Z, Zheng K, Xu X, Zhao F. An automatic segmentation and quantification method for austenite and ferrite phases in duplex stainless steel based on deep learning. Journal of Materials Chemistry A. 2025;13(1):772-85.

[22] Covarrubias-Garcia J, García-Pastor FA, Castelán M. Deep Learning-Based classification of aluminium alloy microstructures using synthetic metallographic data. The International Journal of Advanced Manufacturing Technology. 2025 Nov 8:1-6.

[23] Zhao Q, Kang J, Wu K. Study on the Recognition of Metallurgical Graphs Based on Deep Learning. Metals. 2024 Jun 20;14(6):732.

[24] Müller M, Stiefel M, Bachmann BI, Britz D, Mücklich F. Overview: Machine learning for segmentation and classification of complex steel microstructures. Metals. 2024 May 7;14(5):553.

[25] Wang X, Su H, Li N, Chen Y, Yang Y, Meng H. A Deep Learning Labeling Method for Material Microstructure Image Segmentation. Processes. 2023 Nov 22;11(12):3272.

Ahead of Print Subscription Review Article
Volume 14
03
Received 20/03/2026
Accepted 27/03/2026
Published 11/05/2026
Publication Time 52 Days
33

Login


My IP

PlumX Metrics