Segmentation of pores in carbon fiber reinforced polymers using the U-Net convolutional neural network

This study demonstrates the utilization of deep learning techniques for binary semantic segmentation of pores in carbon fiber reinforced polymers (CFRP) using X-ray computed tomography (XCT) datasets. The proposed workflow is designed to generate efficient segmentation models with reasonable execut...

Full description

Saved in:
Bibliographic Details
Main Authors: Miroslav Yosifov, Patrick Weinberger, Bernhard Plank, Bernhard Fröhler, Markus Hoeglinger, Johann Kastner, Christoph Heinzl
Format: Article
Language:English
Published: Czech Technical University in Prague 2023-10-01
Series:Acta Polytechnica CTU Proceedings
Subjects:
Online Access:https://ojs.cvut.cz/ojs/index.php/APP/article/view/9407
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850059100590702592
author Miroslav Yosifov
Patrick Weinberger
Bernhard Plank
Bernhard Fröhler
Markus Hoeglinger
Johann Kastner
Christoph Heinzl
author_facet Miroslav Yosifov
Patrick Weinberger
Bernhard Plank
Bernhard Fröhler
Markus Hoeglinger
Johann Kastner
Christoph Heinzl
author_sort Miroslav Yosifov
collection DOAJ
description This study demonstrates the utilization of deep learning techniques for binary semantic segmentation of pores in carbon fiber reinforced polymers (CFRP) using X-ray computed tomography (XCT) datasets. The proposed workflow is designed to generate efficient segmentation models with reasonable execution time, applicable even for users using consumer-grade GPU systems. First, U-Net, a convolutional neural network, is modified to handle the segmentation of XCT datasets. In the second step, suitable hyperparameters are determined through a parameter analysis (hyperparameter tuning), and the parameter set with the best result was used for the final training. In the final step, we report on our efforts of implementing the testing stage in open_iA, which allows users to segment datasets with the fully trained model within reasonable time. The model performs well on datasets with both high and low resolution, and even works reasonably for barely visible pores with different shapes and size. In our experiments, we could show that U-Net is suitable for pore segmentation. Despite being trained on a limited number of datasets, it exhibits a satisfactory level of prediction accuracy.
format Article
id doaj-art-ea4bc9edebda41a39a544df7cd4a7d90
institution DOAJ
issn 2336-5382
language English
publishDate 2023-10-01
publisher Czech Technical University in Prague
record_format Article
series Acta Polytechnica CTU Proceedings
spelling doaj-art-ea4bc9edebda41a39a544df7cd4a7d902025-08-20T02:50:59ZengCzech Technical University in PragueActa Polytechnica CTU Proceedings2336-53822023-10-014210.14311/APP.2023.42.0087Segmentation of pores in carbon fiber reinforced polymers using the U-Net convolutional neural networkMiroslav Yosifov0Patrick Weinberger1Bernhard Plank2Bernhard Fröhler3Markus Hoeglinger4Johann Kastner5Christoph Heinzl6University of Antwerp, imec-Visionlab, Department of Physics, Universiteitsplein 1, 2610 Antwerpen, Belgium; University of Applied Sciences Upper Austria, Research Group X-ray Computed Tomography, Stelzhamerstraße 23, 4600 Wels, AustriaUniversity of Applied Sciences Upper Austria, Research Group X-ray Computed Tomography, Stelzhamerstraße 23, 4600 Wels, AustriaUniversity of Applied Sciences Upper Austria, Research Group X-ray Computed Tomography, Stelzhamerstraße 23, 4600 Wels, AustriaUniversity of Applied Sciences Upper Austria, Research Group X-ray Computed Tomography, Stelzhamerstraße 23, 4600 Wels, AustriaUniversity of Applied Sciences Upper Austria, Research Group X-ray Computed Tomography, Stelzhamerstraße 23, 4600 Wels, AustriaUniversity of Applied Sciences Upper Austria, Research Group X-ray Computed Tomography, Stelzhamerstraße 23, 4600 Wels, AustriaUniversity of Passau, Faculty of Computer Science and Mathematics, Innstraße 43, 94032 Passau, Germany This study demonstrates the utilization of deep learning techniques for binary semantic segmentation of pores in carbon fiber reinforced polymers (CFRP) using X-ray computed tomography (XCT) datasets. The proposed workflow is designed to generate efficient segmentation models with reasonable execution time, applicable even for users using consumer-grade GPU systems. First, U-Net, a convolutional neural network, is modified to handle the segmentation of XCT datasets. In the second step, suitable hyperparameters are determined through a parameter analysis (hyperparameter tuning), and the parameter set with the best result was used for the final training. In the final step, we report on our efforts of implementing the testing stage in open_iA, which allows users to segment datasets with the fully trained model within reasonable time. The model performs well on datasets with both high and low resolution, and even works reasonably for barely visible pores with different shapes and size. In our experiments, we could show that U-Net is suitable for pore segmentation. Despite being trained on a limited number of datasets, it exhibits a satisfactory level of prediction accuracy. https://ojs.cvut.cz/ojs/index.php/APP/article/view/9407deep learningsegmentationU-Netcomputed tomographyporescarbon fiber reinforced polymers
spellingShingle Miroslav Yosifov
Patrick Weinberger
Bernhard Plank
Bernhard Fröhler
Markus Hoeglinger
Johann Kastner
Christoph Heinzl
Segmentation of pores in carbon fiber reinforced polymers using the U-Net convolutional neural network
Acta Polytechnica CTU Proceedings
deep learning
segmentation
U-Net
computed tomography
pores
carbon fiber reinforced polymers
title Segmentation of pores in carbon fiber reinforced polymers using the U-Net convolutional neural network
title_full Segmentation of pores in carbon fiber reinforced polymers using the U-Net convolutional neural network
title_fullStr Segmentation of pores in carbon fiber reinforced polymers using the U-Net convolutional neural network
title_full_unstemmed Segmentation of pores in carbon fiber reinforced polymers using the U-Net convolutional neural network
title_short Segmentation of pores in carbon fiber reinforced polymers using the U-Net convolutional neural network
title_sort segmentation of pores in carbon fiber reinforced polymers using the u net convolutional neural network
topic deep learning
segmentation
U-Net
computed tomography
pores
carbon fiber reinforced polymers
url https://ojs.cvut.cz/ojs/index.php/APP/article/view/9407
work_keys_str_mv AT miroslavyosifov segmentationofporesincarbonfiberreinforcedpolymersusingtheunetconvolutionalneuralnetwork
AT patrickweinberger segmentationofporesincarbonfiberreinforcedpolymersusingtheunetconvolutionalneuralnetwork
AT bernhardplank segmentationofporesincarbonfiberreinforcedpolymersusingtheunetconvolutionalneuralnetwork
AT bernhardfrohler segmentationofporesincarbonfiberreinforcedpolymersusingtheunetconvolutionalneuralnetwork
AT markushoeglinger segmentationofporesincarbonfiberreinforcedpolymersusingtheunetconvolutionalneuralnetwork
AT johannkastner segmentationofporesincarbonfiberreinforcedpolymersusingtheunetconvolutionalneuralnetwork
AT christophheinzl segmentationofporesincarbonfiberreinforcedpolymersusingtheunetconvolutionalneuralnetwork