Publication detail

Artificial intelligence in material testing with digital image correlation

ŠČERBA, B. NÁVRAT, T. VAJDÁK, M. ČEPELA, J.

English title

Artificial intelligence in material testing with digital image correlation

Type

article in a collection out of WoS and Scopus

Language

en

Original abstract

Artificial intelligence improves user experience and work efficiency in digital image correlation (DIC) systems by automating certain tasks, thus improving the competitive ability of local DIC system developers in the global market. The main objective is to detect the type of sample used for material testing, choose the proper measurement tool in the DIC system, and find the proper placement of the tool in relation to the sample. This leads to solving problems of computer vision and image processing, such as object localization and classification, using convolutional neural networks. The open-source PyTorch framework is used for machine learning activities. Tasks like data acquisition, selection and labeling, model selection, training and validation, or hyperparameter optimization are dealt with. As a result, about 95 % of the images were detected successfully.

English abstract

Artificial intelligence improves user experience and work efficiency in digital image correlation (DIC) systems by automating certain tasks, thus improving the competitive ability of local DIC system developers in the global market. The main objective is to detect the type of sample used for material testing, choose the proper measurement tool in the DIC system, and find the proper placement of the tool in relation to the sample. This leads to solving problems of computer vision and image processing, such as object localization and classification, using convolutional neural networks. The open-source PyTorch framework is used for machine learning activities. Tasks like data acquisition, selection and labeling, model selection, training and validation, or hyperparameter optimization are dealt with. As a result, about 95 % of the images were detected successfully.

Keywords in English

DIC, digital image correlation, object detection, YOLOv5, convolutional neural network

Released

06.06.2022

Publisher

Czech Technical University in Prague, Faculty of Mechanical Engineering, Department of Mechanics, Biomechanics and Mechatronics, Technická 4, 160 00 Prague 6

Location

Prague

ISBN

978-80-01-07010-9

Book

Experimental Stress Analysis 2022 Book of Extended Abstract

Pages from–to

132–133

Pages count

2

BIBTEX


@inproceedings{BUT182141,
  author="Bořek {Ščerba} and Tomáš {Návrat} and Michal {Vajdák} and Jan {Čepela},
  title="Artificial intelligence in material testing with digital image correlation",
  booktitle="Experimental Stress Analysis 2022 Book of Extended Abstract",
  year="2022",
  month="June",
  pages="132--133",
  publisher="Czech Technical University in Prague, Faculty of Mechanical Engineering, Department of Mechanics, Biomechanics and Mechatronics, Technická 4, 160 00 Prague  6",
  address="Prague",
  isbn="978-80-01-07010-9"
}