Improved YOLOv5l-based Detection of Surface Defects in Hot Rolled Steel Strips

Zhiren Zhu; Liming Zhou; Fankai Chen; Chen Liu; Fanrun Meng

doi:10.54097/fcis.v4i1.9453

Authors

Zhiren Zhu
Liming Zhou
Fankai Chen
Chen Liu
Fanrun Meng

DOI:

https://doi.org/10.54097/fcis.v4i1.9453

Keywords:

Hot Rolled Strip, SimAM Attention Mechanism, C2F Module, Defect Detection

Abstract

To address the problems of complex background, different sizes and easy to miss and mis-detect in the detection of surface defects in hot-rolled strip, an improved YOLOv5l-based method for detecting surface defects in hot-rolled strip is proposed. Firstly, by adding the SimAM attention mechanism module to the aggregation network, the important information is focused with high weights to improve the recall rate of the original algorithm; secondly, by replacing all C3 modules in the YOLOv5l structure with C2F, a richer gradient of information flow is obtained to improve the accuracy rate of the original algorithm. The experimental results show that the average detection accuracy using the improved YOLOv5l improves by 5.3% and the accuracy rate by 8.3% compared to the original network, resulting in higher detection accuracy and lower error and miss detection rates, meeting the requirements of hot-rolled strip steel inspection in industrial manufacturing.

Downloads

Download data is not yet available.

References

Li,Y., et al. "Progress in surface defect detection methods for strip steel." Journal of Iron and Steel Research . doi: 10. 13228/ j. boyuan. issn1001-0963.20220363.

Wang Meng." A multi-scale feature map-based method for detecting defects in strip steel." Digital Technology and Applications 40.04(2022):36-39. doi:10.19695/j.cnki.cn12-1369. 2022.04.12.

Zhang Yan,and Feng Feng." Exploration of strip steel surface defect detection technology." Information and Computer (Theoretical Edition) 33.11(2021):19-22.

Pan Meng, Zhou Deqiang,and Chang Xiang." Characterization of surface defect detection by a novel pulsed leakage magnetic detection method." Sensors and Microsystems 36.12 (2017): 32-35. doi:10.13873/J.1000-9787(2017)12-0032-04.

Wang B, et al. "A new eddy current detection method and its detection effect." Metallurgy of China 31.02(2021):50-54. doi:10.13228/j.boyuan.issn1006-9356.20200350.

Ma,K., et al. "Study on SF_6 decomposition under pin-plate defects based on infrared detection method." High Voltage Electronics 48.12 (2012): 70-74. doi:10.13296/j.1001-1609. hva. 2012.12.015.

Tian W. Research on Curriculum Design Method of Teaching Resource Library based on Deep Learning Technology [C]// Wuhan Zhicheng Times Cultural Proceedings of 6th International Conference on Computer Engineering, Information Science & Application Technology (ICCIA 2023). [publisher unknown], 2023:148-152. doi:10. 26914/ c. cnkihy. 2023. 010734.

Zhang Zemiao, Huo Huan,and Zhao Fengyu." A review of target detection algorithms for deep convolutional neural networks." Small Microcomputer Systems 40.09(2019):1825-1831.

Ning, J., et al. "A review of target detection algorithms for deep learning." Information Recorded Materials 23.10 (2022): 1-4. doi:10.16009/j.cnki.cn13-1295/tq.2022.10.057.

Wen, Xinglin,and Bai, Tao." Design of audio emotion recognition and classification model based on improved multimodal RCNN." Modern Electronics 46.11(2023):114-118. doi:10.16652/j.issn.1004-373x.2023.11.021.

Yang, H. Zhou,and Li, D. Li." YOLOv4 target detection based on improved SPPnet." Electronic Fabrication .22(2021):52-54. doi:10.16589/j.cnki.cn11-3571/tn.2021.22.018.

Wang, Yanqing, et al. "Application of FastRCNN and CNN techniques in the morphological identification of three parasite eggs." Journal of Qiqihar Medical College 43.03(2022):234-237.

Bai Chenshuai, et al. "An improved target detection algorithm based on Faster-RCNN (in English)." Journal of Measurement Science and Instrumentation .

Wang, Dao-Lei, et al. "Improved SSD method for detection of hot spot defects in photovoltaic modules." Journal of Solar Energy 44. 04 (2023):420-425. doi: 10. 19912/ j.0254-0096. tynxb. 2021-1470.

Zhou, Jinwei,and Wang, Jianping." A review of YOLO object detection algorithm research." Journal of Changzhou Institute of Technology 36.01(2023):18-23+88.

Liu, J. Chuan, et al. "Improved RetinaNet for UAV small target detection." Science Technology and Engineering 23.01 (2023): 274-282.

Zhang, Zhengchao." Improving YOLOv5 for lightweight strip steel surface defect detection." Computer Systems Applications . doi:10.15888/j.cnki.csa.009162.

Li, W.G., et al. "Surface defect detection of strip steel based on improved YOLOv3 algorithm." Journal of Electronics 48.07 (2020):1284-1292.

Zou, Wang,and Ji, Chang." An improved YOLOv4-tiny method for real-time detection of strip steel surface defects. " Mechanical Science and Technology . doi:10. 13433/ j. cnki. 1003-8728.20230034.

Li, Shun,and Yang, Ying." Improved YOLOv5-based defect detection for hot-rolled strip steel." Computer Simulation .

Wu Di, et al. "Improved YOLOv5-based surface defect detection in steel." Journal of Shaanxi University of Science and Technology 41.02(2023):162-169. doi: 10. 19481/j. cnki. Issn 2096-398x.2023.02.009.

Zhang, Yang, Liu, Xiaofang,and Li, Wenwei." An improved YOLOv5n-based algorithm for strip steel surface defect detection." Journal of Sichuan University of Light and Chemical Technology (Natural Science Edition) 35.05 (2022): 60-67.

Ma Yanting, et al. "Improved surface defect detection method for strip steel with YOLOv5 network." Journal of Electronic Measurement and Instrumentation 36.08(2022):150-157. doi:10.13382/j.jemi.B2205354.

Dong, Yanhua,and Li, Jia'ao." Research on improving YOLOv5s remote sensing image recognition algorithm." Journal of Jilin Normal University (Natural Science Edition) 44.02 (2023):117-123. doi:10. 16862/ j. cnki. issn1674-3873. 2023. 02.017.

Chuanjun Zhu, et al. "Hybrid defect detection model based on SimAM module and ResNet34 network." Modern Manufacturing Engineering .02(2023):1-9. doi:10.16731/ j. cnki. 1671-3133.2023.02.001.