Machine vision based damage detection for conveyor belt safety using fusion knowledge distillation

Xiaoqiang Guo; Xinhua Liu; Paolo Gardoni; Adam Glowacz; Grzegorz Królczyk; Atilla Incecik; Zhixiong Li

doi:10.1016/j.aej.2023.03.034

Machine vision based damage detection for conveyor belt safety using fusion knowledge distillation

Xiaoqiang Guo, Xinhua Liu^*, Paolo Gardoni, Adam Glowacz, Grzegorz Królczyk, Atilla Incecik, Zhixiong Li

^*Corresponding author for this work

Naval Architecture, Ocean And Marine Engineering

Research output: Contribution to journal › Article › peer-review

18 Citations (Scopus)

37 Downloads (Pure)

Abstract

A belt conveyor system is one of the essential equipment in coal mining. The damages to conveyor belts are hazardous because they would affect the stable operation of a belt conveyor system whilst impairing the coal mining efficiency. To address these problems, a novel conveyor belt damage detection method based on CenterNet is proposed in this paper. The fusion of feature-wise and response-wise knowledge distillation is proposed, which balances the performance and size of the proposed deep neural network. The Fused Channel-Spatial Attention is proposed to compress the latent feature maps efficiently, and the Kullback-Leibler divergence is introduced to minimize the distribution distance between student and teacher networks. Experimental results show that the proposed lightweight object detection model reaches 92.53% mAP and 65.8 FPS. The proposed belt damage detection system can detect conveyor belt damages efficiently and accurately, which indicates its high potential to deploy on end devices.

Original language	English
Pages (from-to)	161-172
Number of pages	12
Journal	Alexandria Engineering Journal
Volume	71
Early online date	25 Mar 2023
DOIs	https://doi.org/10.1016/j.aej.2023.03.034
Publication status	Published - 15 May 2023

Funding

This work is supported by the Key Research and Development Plan (Industrial Foresight and Common Key Technology) for Promoting Scientific and Technological Innovation Project Xuzhou City, Jiangsu Province, China (no. KC19074), the Priority Academic Program Development of Jiangsu Higher Education Institution (PAPD). The research leading to these results has received funding from the Norwegian Financial Mechanism 2014-2021 under Project Contract No 2020/37/K/ST8/02748.

Keywords

belt tear detection
CenterNet
deep learning
reliability and risk

Access to Document

10.1016/j.aej.2023.03.034Licence: CC BY-NC-ND 4.0

Guo-etal-AEJ-2023-Machine-vision-based-damage-detection-for-conveyorFinal published version, 2.41 MBLicence: CC BY-NC-ND 4.0

Cite this

@article{a4a190db1e1042a3a611bf588726fabc,

title = "Machine vision based damage detection for conveyor belt safety using fusion knowledge distillation",

abstract = "A belt conveyor system is one of the essential equipment in coal mining. The damages to conveyor belts are hazardous because they would affect the stable operation of a belt conveyor system whilst impairing the coal mining efficiency. To address these problems, a novel conveyor belt damage detection method based on CenterNet is proposed in this paper. The fusion of feature-wise and response-wise knowledge distillation is proposed, which balances the performance and size of the proposed deep neural network. The Fused Channel-Spatial Attention is proposed to compress the latent feature maps efficiently, and the Kullback-Leibler divergence is introduced to minimize the distribution distance between student and teacher networks. Experimental results show that the proposed lightweight object detection model reaches 92.53% mAP and 65.8 FPS. The proposed belt damage detection system can detect conveyor belt damages efficiently and accurately, which indicates its high potential to deploy on end devices.",

keywords = "belt tear detection, CenterNet, deep learning, reliability and risk",

author = "Xiaoqiang Guo and Xinhua Liu and Paolo Gardoni and Adam Glowacz and Grzegorz Kr{\'o}lczyk and Atilla Incecik and Zhixiong Li",

year = "2023",

month = may,

day = "15",

doi = "10.1016/j.aej.2023.03.034",

language = "English",

volume = "71",

pages = "161--172",

}

TY - JOUR

T1 - Machine vision based damage detection for conveyor belt safety using fusion knowledge distillation

AU - Guo, Xiaoqiang

AU - Liu, Xinhua

AU - Gardoni, Paolo

AU - Glowacz, Adam

AU - Królczyk, Grzegorz

AU - Incecik, Atilla

AU - Li, Zhixiong

PY - 2023/5/15

Y1 - 2023/5/15

N2 - A belt conveyor system is one of the essential equipment in coal mining. The damages to conveyor belts are hazardous because they would affect the stable operation of a belt conveyor system whilst impairing the coal mining efficiency. To address these problems, a novel conveyor belt damage detection method based on CenterNet is proposed in this paper. The fusion of feature-wise and response-wise knowledge distillation is proposed, which balances the performance and size of the proposed deep neural network. The Fused Channel-Spatial Attention is proposed to compress the latent feature maps efficiently, and the Kullback-Leibler divergence is introduced to minimize the distribution distance between student and teacher networks. Experimental results show that the proposed lightweight object detection model reaches 92.53% mAP and 65.8 FPS. The proposed belt damage detection system can detect conveyor belt damages efficiently and accurately, which indicates its high potential to deploy on end devices.

AB - A belt conveyor system is one of the essential equipment in coal mining. The damages to conveyor belts are hazardous because they would affect the stable operation of a belt conveyor system whilst impairing the coal mining efficiency. To address these problems, a novel conveyor belt damage detection method based on CenterNet is proposed in this paper. The fusion of feature-wise and response-wise knowledge distillation is proposed, which balances the performance and size of the proposed deep neural network. The Fused Channel-Spatial Attention is proposed to compress the latent feature maps efficiently, and the Kullback-Leibler divergence is introduced to minimize the distribution distance between student and teacher networks. Experimental results show that the proposed lightweight object detection model reaches 92.53% mAP and 65.8 FPS. The proposed belt damage detection system can detect conveyor belt damages efficiently and accurately, which indicates its high potential to deploy on end devices.

KW - belt tear detection

KW - CenterNet

KW - deep learning

KW - reliability and risk

U2 - 10.1016/j.aej.2023.03.034

DO - 10.1016/j.aej.2023.03.034

M3 - Article

AN - SCOPUS:85150927171

SN - 1110-0168

VL - 71

SP - 161

EP - 172

JO - Alexandria Engineering Journal

JF - Alexandria Engineering Journal

ER -

Machine vision based damage detection for conveyor belt safety using fusion knowledge distillation

Abstract

Funding

Keywords

Access to Document

Fingerprint

Cite this