Using deep learning for defect classification on a small weld X-ray image dataset
Ver/
Compartir
Estadísticas
Ver Estadísticas de usoMetadatos
Mostrar el registro completo del ítemAutor
Ajmi, Chiraz; Zapata Pérez, Juan Francisco; Martínez Álvarez, José Javier; Doménech Asensi, Ginés; Ruiz Merino, Ramón JesúsGrupo de investigación
Desarrollo de sistemas y circuitos electrónicos y microelectrónicosÁrea de conocimiento
Ingeniería TelemáticaPatrocinadores
The authors wants to acknowledge the work of the rest of the participants in this project, namely: J.A. López-Alcantud, P. Rubio-Ibañez, Universidad Politécnica de Cartagena, J.A. Díaz-Madrid, Centro Universitario de la Defensa - UPCT and T.J. Kazmierski, University of Southampton. This work has been partially funded by Spanish government through project numbered RTI2018-097088-B-C33 (MINECO/FEDER,UE).Realizado en/con
University of TunisFecha de publicación
2020-09-04Editorial
Springer ScienceCita bibliográfica
Ajmi, C., Zapata, J., Martínez-Álvarez, J.J. et al. Using Deep Learning for Defect Classification on a Small Weld X-ray Image Dataset. J Nondestruct Eval 39, 68 (2020). https://doi.org/10.1007/s10921-020-00719-9Revisión por pares
siPalabras clave
Industrial X-ray ImagesWelding defects
Heterogeneities classification
Deep learning
Machine learning
Resumen
This document provides a comparative evaluation of the performance of a deep learning network for different combinations of parameters and hyper-parameters. Although there are numerous studies that report on performance in deep learning networks for ordinary data sets, their performance on small data sets is much less evaluated. The objective of this work is to demonstrate that such a challenging small data set, such as a welding X-ray image data set, can be trained and evaluated obtaining high precision and that it is possible thanks to data augmentation. In fact, this article shows that data augmentation, also a typical technique in any learning process on a large data set, plus that two image channels, such as channels B (blue) and G (green), both are replaced by the Canny edge map and a binary image provided by an adaptive Gaussian threshold, respectively, gives to the network a 3% increase in accuracy, approximately. In summary, the objective of this work is to present the methodology ...
Colecciones
- Artículos [1713]
El ítem tiene asociados los siguientes ficheros de licencia:
Redes sociales