A CNN-Based Method for Quantitative Assessment of Steel Microstructures in Welded Zones

The mechanical performance of metallic components is intrinsically linked to their microstructural features. However, the manual quantification of microconstituents in metallographic images remains a time-consuming and subjective task, often requiring over 15 min per image by a trained expert. To ad...

Full description

Saved in:
Bibliographic Details
Main Authors: Cássio Danelon de Almeida, Thales Tozatto Filgueiras, Moisés Luiz Lagares, Bruno da Silva Macêdo, Camila Martins Saporetti, Matteo Bodini, Leonardo Goliatt
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Fibers
Subjects:
Online Access:https://www.mdpi.com/2079-6439/13/5/66
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849711856003842048
author Cássio Danelon de Almeida
Thales Tozatto Filgueiras
Moisés Luiz Lagares
Bruno da Silva Macêdo
Camila Martins Saporetti
Matteo Bodini
Leonardo Goliatt
author_facet Cássio Danelon de Almeida
Thales Tozatto Filgueiras
Moisés Luiz Lagares
Bruno da Silva Macêdo
Camila Martins Saporetti
Matteo Bodini
Leonardo Goliatt
author_sort Cássio Danelon de Almeida
collection DOAJ
description The mechanical performance of metallic components is intrinsically linked to their microstructural features. However, the manual quantification of microconstituents in metallographic images remains a time-consuming and subjective task, often requiring over 15 min per image by a trained expert. To address this limitation, this study proposes an automated approach for quantifying the microstructural constituents from low-carbon steel welded zone images using convolutional neural networks (CNNs). A dataset of 210 micrographs was expanded to 720 samples through data augmentation to improve model generalization. Two architectures (AlexNet and VGG16) were trained from scratch, while three pre-trained models (VGG19, InceptionV3, and Xception) were fine-tuned. Among these, VGG19 optimized with stochastic gradient descent (SGD) achieved the best predictive performance, with an <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mi>R</mi><mn>2</mn></msup></semantics></math></inline-formula> of 0.838, MAE of 5.01%, and RMSE of 6.88%. The results confirm the effectiveness of CNNs for reliable and efficient microstructure quantification, offering a significant contribution to computational metallography.
format Article
id doaj-art-e2c6244d9ef1469eae1f4fcad993312b
institution DOAJ
issn 2079-6439
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Fibers
spelling doaj-art-e2c6244d9ef1469eae1f4fcad993312b2025-08-20T03:14:31ZengMDPI AGFibers2079-64392025-05-011356610.3390/fib13050066A CNN-Based Method for Quantitative Assessment of Steel Microstructures in Welded ZonesCássio Danelon de Almeida0Thales Tozatto Filgueiras1Moisés Luiz Lagares2Bruno da Silva Macêdo3Camila Martins Saporetti4Matteo Bodini5Leonardo Goliatt6Computational Modeling Program, Federal University of Juiz de Fora, Juiz de Fora 36036-900, MG, BrazilDepartment of Electrical Engineering, Federal University of Juiz de Fora, Juiz de Fora 36036-900, MG, BrazilDepartment of Mechanical Engineering, Federal University of Juiz de Fora, Juiz de Fora 36036-900, MG, BrazilDepartment of Computer Science, Federal University of Lavras, Lavras 37200-00, MG, BrazilDepartment of Computational Modeling, Polytechnic Institute, Rio de Janeiro State University, Nova Friburgo 22000-900, RJ, BrazilDipartimento di Economia, Management e Metodi Quantitativi, Università degli Studi di Milano, Via Conservatorio 7, 20122 Milano, ItalyDepartment of Computational and Applied Mechanics, Federal University of Juiz de Fora, Juiz de Fora 36036-900, MG, BrazilThe mechanical performance of metallic components is intrinsically linked to their microstructural features. However, the manual quantification of microconstituents in metallographic images remains a time-consuming and subjective task, often requiring over 15 min per image by a trained expert. To address this limitation, this study proposes an automated approach for quantifying the microstructural constituents from low-carbon steel welded zone images using convolutional neural networks (CNNs). A dataset of 210 micrographs was expanded to 720 samples through data augmentation to improve model generalization. Two architectures (AlexNet and VGG16) were trained from scratch, while three pre-trained models (VGG19, InceptionV3, and Xception) were fine-tuned. Among these, VGG19 optimized with stochastic gradient descent (SGD) achieved the best predictive performance, with an <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><msup><mi>R</mi><mn>2</mn></msup></semantics></math></inline-formula> of 0.838, MAE of 5.01%, and RMSE of 6.88%. The results confirm the effectiveness of CNNs for reliable and efficient microstructure quantification, offering a significant contribution to computational metallography.https://www.mdpi.com/2079-6439/13/5/66convolutional neural networksmicrostructure quantificationlow-carbon steeldeep learningimage analysis
spellingShingle Cássio Danelon de Almeida
Thales Tozatto Filgueiras
Moisés Luiz Lagares
Bruno da Silva Macêdo
Camila Martins Saporetti
Matteo Bodini
Leonardo Goliatt
A CNN-Based Method for Quantitative Assessment of Steel Microstructures in Welded Zones
Fibers
convolutional neural networks
microstructure quantification
low-carbon steel
deep learning
image analysis
title A CNN-Based Method for Quantitative Assessment of Steel Microstructures in Welded Zones
title_full A CNN-Based Method for Quantitative Assessment of Steel Microstructures in Welded Zones
title_fullStr A CNN-Based Method for Quantitative Assessment of Steel Microstructures in Welded Zones
title_full_unstemmed A CNN-Based Method for Quantitative Assessment of Steel Microstructures in Welded Zones
title_short A CNN-Based Method for Quantitative Assessment of Steel Microstructures in Welded Zones
title_sort cnn based method for quantitative assessment of steel microstructures in welded zones
topic convolutional neural networks
microstructure quantification
low-carbon steel
deep learning
image analysis
url https://www.mdpi.com/2079-6439/13/5/66
work_keys_str_mv AT cassiodanelondealmeida acnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT thalestozattofilgueiras acnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT moisesluizlagares acnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT brunodasilvamacedo acnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT camilamartinssaporetti acnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT matteobodini acnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT leonardogoliatt acnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT cassiodanelondealmeida cnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT thalestozattofilgueiras cnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT moisesluizlagares cnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT brunodasilvamacedo cnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT camilamartinssaporetti cnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT matteobodini cnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones
AT leonardogoliatt cnnbasedmethodforquantitativeassessmentofsteelmicrostructuresinweldedzones