Generation of High-Precision Ground Penetrating Radar Images Using Improved Least Square Generative Adversarial Networks

Deep learning models have achieved success in image recognition and have shown great potential for interpretation of ground penetrating radar (GPR) data. However, training reliable deep learning models requires massive labeled data, which are usually not easy to obtain due to the high costs of data...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Yunpeng Yue, Hai Liu, Xu Meng, Yinguang Li, Yanliang Du
Formato:	article
Lenguaje:	EN
Publicado:	MDPI AG 2021
Materias:	ground penetrating radar (GPR) deep learning least square generative adversarial networks (LSGAN) data augmentation Science Q
Acceso en línea:	https://doaj.org/article/7085fe613529485290866780cf1640ea
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:7085fe613529485290866780cf1640ea
record_format	dspace
spelling	oai:doaj.org-article:7085fe613529485290866780cf1640ea2021-11-25T18:54:34ZGeneration of High-Precision Ground Penetrating Radar Images Using Improved Least Square Generative Adversarial Networks10.3390/rs132245902072-4292https://doaj.org/article/7085fe613529485290866780cf1640ea2021-11-01T00:00:00Zhttps://www.mdpi.com/2072-4292/13/22/4590https://doaj.org/toc/2072-4292Deep learning models have achieved success in image recognition and have shown great potential for interpretation of ground penetrating radar (GPR) data. However, training reliable deep learning models requires massive labeled data, which are usually not easy to obtain due to the high costs of data acquisition and field validation. This paper proposes an improved least square generative adversarial networks (LSGAN) model which employs the loss functions of LSGAN and convolutional neural networks (CNN) to generate GPR images. This model can generate high-precision GPR data to address the scarcity of labelled GPR data. We evaluate the proposed model using Frechet Inception Distance (FID) evaluation index and compare it with other existing GAN models and find it outperforms the other two models on a lower FID score. In addition, the adaptability of the LSGAN-generated images for GPR data augmentation is investigated by YOLOv4 model, which is employed to detect rebars in field GPR images. It is verified that inclusion of LSGAN-generated images in the training GPR dataset can increase the target diversity and improve the detection precision by 10%, compared with the model trained on the dataset containing 500 field GPR images.Yunpeng YueHai LiuXu MengYinguang LiYanliang DuMDPI AGarticleground penetrating radar (GPR)deep learningleast square generative adversarial networks (LSGAN)data augmentationScienceQENRemote Sensing, Vol 13, Iss 4590, p 4590 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	ground penetrating radar (GPR) deep learning least square generative adversarial networks (LSGAN) data augmentation Science Q
spellingShingle	ground penetrating radar (GPR) deep learning least square generative adversarial networks (LSGAN) data augmentation Science Q Yunpeng Yue Hai Liu Xu Meng Yinguang Li Yanliang Du Generation of High-Precision Ground Penetrating Radar Images Using Improved Least Square Generative Adversarial Networks
description	Deep learning models have achieved success in image recognition and have shown great potential for interpretation of ground penetrating radar (GPR) data. However, training reliable deep learning models requires massive labeled data, which are usually not easy to obtain due to the high costs of data acquisition and field validation. This paper proposes an improved least square generative adversarial networks (LSGAN) model which employs the loss functions of LSGAN and convolutional neural networks (CNN) to generate GPR images. This model can generate high-precision GPR data to address the scarcity of labelled GPR data. We evaluate the proposed model using Frechet Inception Distance (FID) evaluation index and compare it with other existing GAN models and find it outperforms the other two models on a lower FID score. In addition, the adaptability of the LSGAN-generated images for GPR data augmentation is investigated by YOLOv4 model, which is employed to detect rebars in field GPR images. It is verified that inclusion of LSGAN-generated images in the training GPR dataset can increase the target diversity and improve the detection precision by 10%, compared with the model trained on the dataset containing 500 field GPR images.
format	article
author	Yunpeng Yue Hai Liu Xu Meng Yinguang Li Yanliang Du
author_facet	Yunpeng Yue Hai Liu Xu Meng Yinguang Li Yanliang Du
author_sort	Yunpeng Yue
title	Generation of High-Precision Ground Penetrating Radar Images Using Improved Least Square Generative Adversarial Networks
title_short	Generation of High-Precision Ground Penetrating Radar Images Using Improved Least Square Generative Adversarial Networks
title_full	Generation of High-Precision Ground Penetrating Radar Images Using Improved Least Square Generative Adversarial Networks
title_fullStr	Generation of High-Precision Ground Penetrating Radar Images Using Improved Least Square Generative Adversarial Networks
title_full_unstemmed	Generation of High-Precision Ground Penetrating Radar Images Using Improved Least Square Generative Adversarial Networks
title_sort	generation of high-precision ground penetrating radar images using improved least square generative adversarial networks
publisher	MDPI AG
publishDate	2021
url	https://doaj.org/article/7085fe613529485290866780cf1640ea
work_keys_str_mv	AT yunpengyue generationofhighprecisiongroundpenetratingradarimagesusingimprovedleastsquaregenerativeadversarialnetworks AT hailiu generationofhighprecisiongroundpenetratingradarimagesusingimprovedleastsquaregenerativeadversarialnetworks AT xumeng generationofhighprecisiongroundpenetratingradarimagesusingimprovedleastsquaregenerativeadversarialnetworks AT yinguangli generationofhighprecisiongroundpenetratingradarimagesusingimprovedleastsquaregenerativeadversarialnetworks AT yanliangdu generationofhighprecisiongroundpenetratingradarimagesusingimprovedleastsquaregenerativeadversarialnetworks
_version_	1718410594933014528

Generation of High-Precision Ground Penetrating Radar Images Using Improved Least Square Generative Adversarial Networks

Ejemplares similares