钢材表面缺陷检测的YOLOv5s算法优化研究

doi:10.3778/j.issn.1002-8331.2307-0275

摘要/Abstract

摘要： 针对YOLOv5对钢材缺陷复杂特征提取能力不足且检测结果易受背景环境干扰等问题，提出一种基于YOLOv5s的钢材表面缺陷检测算法。该算法在Backbone的核心特征提取模块C3中引入CBAM注意力，提升Backbone层对于关键信息的关注度；利用CARAFE替换最近邻插值算法，降低了上采样操作对于特征信息造成的损失；提出用融合跨阶段局部网络的金字塔池化结构SPPCPSC替换YOLOv5中的SPPF，提升网络的表达能力和感知能力。实验结果表明，提出YOLOv5s改进模型在NEU-DET数据集上的mAP@0.5达到了76.6%，比YOLOv5s提升2.3个百分点，模型参数量与基线模型基本一致，而CARAFE是导致改进模型检测速度降低的主要原因。除此，实验结果还发现CARAFE与SPPCSPC_group组合使用，对于模型的检测准确度有良好的提升作用。

关键词: YOLOv5, 注意力机制, 跨阶段局部通道金字塔池化结构（SPPCSPC）, 特征提取, 缺陷检测

Abstract: Aiming at the problems that YOLOv5 has insufficient ability to extract complex features of steel defects and the detect results are susceptible to background environment, a steel surface defect detection algorithm based on YOLOv5s is proposed. This algorithm introduces CBAM attention into C3 to enhance attention to key information. It utilizes the CARAFE to replace the nearest neighbor interpolation algorithm, reducing the loss of feature information caused by upsampling. It proposes to replace the SPPF in YOLOv5 with the SPPCPSC, which can improve the expressive ability of the network. The experimental results show that the mAP@0.5 of the proposed YOLOv5s improved model on the NEU-DET dataset reaches 76.6%, which is 2.3 percentage points higher than that of YOLOv5s. The model parameters are basically the same as YOLOv5s. The CARAFE module is the main reason for the slowdown of the improved model detection speed. In addition, the combination of the CARAFE and the SPPCSPC_group has a good effect on the detection accuracy of the model.

Key words: YOLOv5, attention mechanism, spatial pyramid pooling-cross stage partial channel (SPPCSPC), feature extraction, defect detection

徐洪俊, 唐自强, 张锦东, 朱沛华. 钢材表面缺陷检测的YOLOv5s算法优化研究[J]. 计算机工程与应用, 2024, 60(7): 306-314.

XU Hongjun, TANG Ziqiang, ZHANG Jindong, ZHU Peihua. Research on Optimization of YOLOv5s Detection Algorithm for Steel Surface Defect[J]. Computer Engineering and Applications, 2024, 60(7): 306-314.

参考文献

[1] FU G Z, SUN P Z, ZHU W B, et al. A deep-learning-based approach for fast and robust steel surface defects classification[J]. Optics and Lasers in Engineering, 2019, 121: 397-405.
[2] LV X M, DUAN F J, JIANG J J, et al. Deep metallic surface defect detection: the new benchmark and detection network[J]. Sensors: Basel, Switzerland, 2020, 20(6): 1562.
[3] VANNOCCI M, RITACCO A, CASTELLANO A, et al. Flatness defect detection and classification in hot rolled steel strips using convolutional neural networks[M]//Advances in computational intelligence. Cham: Springer International Publishing, 2019: 220-234.
[4] HAN C J, LI G Y, LIU Z. Two-stage edge reuse network for salient object detection of strip steel surface defects[J]. IEEE Transactions on Instrumentation and Measurement, 2022, 71: 1-12.
[5] 王海云, 王剑平, 罗付华. 融合多层次特征Faster R-CNN的金属板带材表面缺陷检测研究[J]. 机械科学与技术, 2021, 40(2): 262-269.
WANG H Y, WANG J P, LUO F H. Study on surface defect detection of metal sheet and strip using faster R-CNN with multilevel feature[J]. Mechanical Science and Technology for Aerospace Engineering, 2021, 40(2): 262-269.
[6] LI J Y, SU Z F, GENG J H, et al. Real-time detection of steel strip surface defects based on improved YOLO detection network[J]. IFAC PapersOnLine, 2018, 51(21): 76-81.
[7] 程婧怡, 段先华, 朱伟. 改进YOLOv3的金属表面缺陷检测研究[J]. 计算机工程与应用, 2021, 57(19): 252-258.
CHEN J Y, DUAN X H, ZHU W. Research on metal surface defect detection by improved YOLOv3[J]. Computer Engineering and Applications, 2021, 57(19): 252-258.
[8] KOU X P, LIU S J, CHENG K Q, et al. Development of a YOLO-V3-based model for detecting defects on steel strip surface[J]. Measurement, 2021, 182: 109454.
[9] 曹义亲, 伍铭林, 徐露. 基于改进YOLOv5算法的钢材表面缺陷检测[J]. 图学学报, 2023, 44(2): 335-345.
CAO Y Q, WU M L, XU L. Steel surface defect detection based on improved YOLOv5 algorithm[J]. Journal of Graphics, 2023, 44(2): 335-345.
[10] 卢海滨. 基于YOLOv5s的钢材表面缺陷检测研究[D]. 大庆: 东北石油大学, 2023.
LU H B. Research on steel surface defect detection based on YOLOv5s[D]. Daqing: Northeast Petroleum University, 2023.
[11] 司宏翔, 刘国巍. 针对小目标缺陷的YOLOv5s锚框改进方法研究[J]. 无线互联科技, 2023, 19(2): 125-127.
SI H X, LIU G W. Research on improvement method of YOLOv5s anchor frame for small target defects[J]. Wireless Internet Technology. 2023, 19(2): 125-127.
[12] 谷长江, 高法钦. 改进YOLOv5s的钢材表面缺陷检测[J]. 软件工程, 2023, 26(8): 31-34.
GU C J, GAO F Q. Steel surface defect detection based on improved YOLOv5s[J]. Software Engineering, 2023, 26(8): 31-34.
[13] 吴敌, 李明辉, 马文凯, 等. 基于改进YOLOv5的钢材表面缺陷检测[J]. 陕西科技大学学报, 2023, 41(2): 162-169.
WU D, LI M H, MA W K, et al. Surface defect detection of steel based on improved YOLOv5[J]. Journal of Shaanxi University of Science & Technology, 2023, 41(2): 162-169.
[14] LIN T Y, DOLLáR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2017: 936-944.
[15] LIU S, QI L, QIN H F, et al. Path aggregation network for instance segmentation[C]//2018 IEEE CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2018: 8759-8768.
[16] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[J]. arXiv:1807.06521, 2018.
[17] 段必冲, 马明涛. 基于改进YOLOv5算法的口罩检测研究[J]. 计算机工程与应用, 2023, 59(16): 223-231.
DUAN B C, MA M T. Research on improved mask detection method based on YOLOv5 algorithm[J]. Computer Engineering and Applications , 2023, 59(16): 223-231.
[18] WANG J Q, CHEN K, XU R, et al. CARAFE: content-aware reassembly of features[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019: 3007-3016.
[19] 邵延华, 张铎, 楚红雨, 等. 基于深度学习的YOLO目标检测综述[J]. 电子与信息学报, 2022, 44(10): 3697-3708.
SHAO Y H, ZHANG D, CHU H Y, et al. A review of YOLO object detection based on deep learning[J]. Journal of Electronics & Information Technology, 2022, 44(10): 3697-3708.
[20] WANG C, MARK L. CSPNet: a new backbone that can enhance learning capability of CNN[J]. arXiv:2003.09119, 2020.
[21] 齐向明, 柴蕊, 高一萌. 重构SPPCSPC与优化下采样的小目标检测算法[J]. 计算机工程与应用, 2023, 59(20): 158-166.
QI X M, CHAI R, GAO Y M. Algorithm of reconstructed SPPCSPC and optimized downsampling for small object detection[J]. Computer Engineering and Applications, 2023, 59(20): 158-166.
[22] 何敏, 秦亮, 赵峰, 等. 面向电力系统现场作业的安全风险管控智能检测算法[J]. 高电压技术, 2023, 49(6): 2442-2457.
HE M, QIN L, ZHAO F, et al. Intelligent detection algorithm of security risk management and control for power system on-site operation[J]. High Voltage Engineering, 2023, 49(6): 2442-2457.
[23] 贾天豪, 彭力, 戴菲菲. 引入残差学习与多尺度特征增强的目标检测器[J]. 计算机科学与探索, 2023, 17(5): 1102-1111.
JIA T H, PENG L, DAI F F. Object detector with residual learning and multi-scale feature enhancement[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(5): 1102-1111.
[24] 王鑫, 张铭. 基于申威众核架构的分组卷积计算加速与优化[J]. 计算机应用研究, 2023, 40(6): 1745-1749.
WANG X, ZHANG M. Acceleration and optimization of group convolution calculation based on SW many-core architecture[J]. Application Research of Computers, 2023, 40(6): 1745-1749.
[25] HOU Q B, ZHOU D Q, FENG J S. Coordinate attention for efficient mobile network design[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, 2021: 13713-13722.
[26] HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 7132-7141.
[27] 肖鹏程, 徐文广, 张妍, 等. 基于SE注意力机制的废钢分类评级方法[J]. 工程科学学报, 2023, 45(8): 1342-1352.
XIAO P C, XU W G, ZHANG Y, et al. Research on scrap classification and rating method based on SE attention mechanism[J]. Chinese Journal of Engineering, 2023, 45(8): 1342-1352.
[28] 李孟浩, 袁三男. 改进YOLOv5s的交通标识检测算法[J]. 南京信息工程大学学报(自然科学版), 2024, 16(1): 11-19.
LI M H, YUAN S N. Traffic sign detection algorithm based on improved YOLOv5s[J]. Journal of Nanjing University of Information Science & Technology (Natural Science Edition), 2024, 16(1): 11-19.