Automatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional Networks

Sorting gangue from raw coal is an essential concern in coal mining engineering. Prior to separation, the location and shape of the gangue should be extracted from the raw coal image. Several approaches regarding automatic detection of gangue have been proposed to date; however, none of them is sati...

Full description

Bibliographic Details
Main Authors:	Rong Gao, Zhaoyun Sun, Wei Li, Lili Pei, Yuanjiao Hu, Liyang Xiao
Format:	Article
Language:	English
Published:	MDPI AG 2020-02-01
Series:	Energies
Subjects:	coal and gangue u-net feature extraction segmentation
Online Access:	https://www.mdpi.com/1996-1073/13/4/829

id	doaj-df2ddb9238ed4f83b8138bb9e4d4e179
record_format	Article
spelling	doaj-df2ddb9238ed4f83b8138bb9e4d4e1792020-11-25T02:16:09ZengMDPI AGEnergies1996-10732020-02-0113482910.3390/en13040829en13040829Automatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional NetworksRong Gao0Zhaoyun Sun1Wei Li2Lili Pei3Yuanjiao Hu4Liyang Xiao5School of Information Engineering, Chang’an University, Xi’an 710064, ChinaSchool of Information Engineering, Chang’an University, Xi’an 710064, ChinaSchool of Information Engineering, Chang’an University, Xi’an 710064, ChinaSchool of Information Engineering, Chang’an University, Xi’an 710064, ChinaSchool of Information Engineering, Chang’an University, Xi’an 710064, ChinaSchool of Information Engineering, Chang’an University, Xi’an 710064, ChinaSorting gangue from raw coal is an essential concern in coal mining engineering. Prior to separation, the location and shape of the gangue should be extracted from the raw coal image. Several approaches regarding automatic detection of gangue have been proposed to date; however, none of them is satisfying. Therefore, this paper aims to conduct gangue segmentation using a U-shape fully convolutional neural network (U-Net). The proposed network is trained to segment gangue from raw coal images collected under complex environmental conditions. The probability map outputted by the network was used to obtain the location and shape information of gangue. The proposed solution was trained on a dataset consisting of 54 shortwave infrared (SWIR) raw coal images collected from Datong Coalfield. The performance of the network was tested with six never seen images, achieving an average area under the receiver operating characteristics (AUROC) value of 0.96. The resulting intersection over union (IoU) was on average equal to 0.86. The results show the potential of using deep learning methods to perform gangue segmentation under various conditions.https://www.mdpi.com/1996-1073/13/4/829coal and gangueu-netfeature extractionsegmentation
collection	DOAJ
language	English
format	Article
sources	DOAJ
author	Rong Gao Zhaoyun Sun Wei Li Lili Pei Yuanjiao Hu Liyang Xiao
spellingShingle	Rong Gao Zhaoyun Sun Wei Li Lili Pei Yuanjiao Hu Liyang Xiao Automatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional Networks Energies coal and gangue u-net feature extraction segmentation
author_facet	Rong Gao Zhaoyun Sun Wei Li Lili Pei Yuanjiao Hu Liyang Xiao
author_sort	Rong Gao
title	Automatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional Networks
title_short	Automatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional Networks
title_full	Automatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional Networks
title_fullStr	Automatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional Networks
title_full_unstemmed	Automatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional Networks
title_sort	automatic coal and gangue segmentation using u-net based fully convolutional networks
publisher	MDPI AG
series	Energies
issn	1996-1073
publishDate	2020-02-01
description	Sorting gangue from raw coal is an essential concern in coal mining engineering. Prior to separation, the location and shape of the gangue should be extracted from the raw coal image. Several approaches regarding automatic detection of gangue have been proposed to date; however, none of them is satisfying. Therefore, this paper aims to conduct gangue segmentation using a U-shape fully convolutional neural network (U-Net). The proposed network is trained to segment gangue from raw coal images collected under complex environmental conditions. The probability map outputted by the network was used to obtain the location and shape information of gangue. The proposed solution was trained on a dataset consisting of 54 shortwave infrared (SWIR) raw coal images collected from Datong Coalfield. The performance of the network was tested with six never seen images, achieving an average area under the receiver operating characteristics (AUROC) value of 0.96. The resulting intersection over union (IoU) was on average equal to 0.86. The results show the potential of using deep learning methods to perform gangue segmentation under various conditions.
topic	coal and gangue u-net feature extraction segmentation
url	https://www.mdpi.com/1996-1073/13/4/829
work_keys_str_mv	AT ronggao automaticcoalandganguesegmentationusingunetbasedfullyconvolutionalnetworks AT zhaoyunsun automaticcoalandganguesegmentationusingunetbasedfullyconvolutionalnetworks AT weili automaticcoalandganguesegmentationusingunetbasedfullyconvolutionalnetworks AT lilipei automaticcoalandganguesegmentationusingunetbasedfullyconvolutionalnetworks AT yuanjiaohu automaticcoalandganguesegmentationusingunetbasedfullyconvolutionalnetworks AT liyangxiao automaticcoalandganguesegmentationusingunetbasedfullyconvolutionalnetworks
_version_	1724892374888349696

Automatic Coal and Gangue Segmentation Using U-Net Based Fully Convolutional Networks

Similar Items