基于SBSI-YOLO11的轻量化轴承外观缺陷检测算法

doi:10.16088/j.issn.1001-6600.2024121901

摘要/Abstract

摘要： 针对现有深度学习模型在轴承外观缺陷检测中存在的检测精度低、模型参数大以及小目标检测困难等问题,本文提出一种轻量化的检测算法SBSI-YOLO11。该算法以YOLO11n为基础网络架构,首先,在主干网络中引入SPD-Conv(space-to-depth convolution)模块,降低特征图的分辨率,增强特征提取并降低模型的参数量;其次,在Neck部分引入BiFPN(bidirectional feature pyramid network)双向特征金字塔网络以及注意力机制SGE(spatial group-wise enhance),提高模型对小目标的检测性能;最后,引入Inner-CIoU损失函数,提高模型的定位能力。实验结果表明,与基础模型相比,SBSI-YOLO11模型展示出良好的综合检测性能,mAP达到90.4%,提高2.9个百分点,参数量降低11.5%,运算量降低12.7%,能够较好适应当前工业现场对轴承外观缺陷的检测需求。

关键词: 轴承缺陷检测, 小目标检测, YOLO11, 轻量化, 深度学习

Abstract: In order to solve the problems of low detection accuracy, large model parameters and difficult detection of small targets in the detection of bearing appearance defects in the existing deep learning models, a lightweight detection algorithm SBSI-YOLO11 is proposed. Firstly, the SPD-Conv (space-to-depth convolution) module is introduced into the backbone network to reduce the resolution of the feature map, enhance the feature extraction and reduce the number of parameters of the model. Secondly, the Bidirectional Feature Pyramid Network (BiFPN) and the Attention Mechanism (SGE) are introduced in the Neck part to improve the detection performance of the model for small targets. Finally, the Inner-CIoU loss function is introduced to improve the positioning ability of the model. Experimental results show that compared with the basic model, the SBSI-YOLO11 model shows good comprehensive detection performance, with an mAP of 90.4%, an increase of 2.9 percentage points, a decrease of 11.5% in the amount of parameters, and a decrease of 12.7% in the amount of calculations, which can better meet the detection needs of the current industrial site for bearing appearance defects.

Key words: bearing defect detection, small target detection, YOLO11, lightweight, deep learning

中图分类号: TP391.41

魏梓书, 陈志刚, 王衍学, 哈斯铁尔·马德提汗. 基于SBSI-YOLO11的轻量化轴承外观缺陷检测算法[J]. 广西师范大学学报（自然科学版）, 2025, 43(6): 80-91.

WEI Zishu, CHEN Zhigang, WANG Yanxue, Hasitieer Madetihan. Lightweight Bearing Defect Detection Algorithm Based on SBSI-YOLO11[J]. Journal of Guangxi Normal University(Natural Science Edition), 2025, 43(6): 80-91.

参考文献

[1] 宋向金, 赵文祥. 交流电机信号特征分析的滚动轴承故障诊断方法综述[J]. 中国电机工程学报, 2022, 42(4): 1582-1596. DOI: 10.13334/j.0258-8013.pcsee.210760.
[2] 姚景丽, 程光, 万飞, 等. 改进YOLOv8的轻量化轴承缺陷检测算法[J]. 计算机工程与应用, 2024, 60(21): 205-214. DOI: 10.3778/j.issn.1002-8331.2403-0444.
[3] 张驰, 郭媛, 黎明. 人工神经网络模型发展及应用综述[J]. 计算机工程与应用, 2021, 57(11): 57-69. DOI: 10.3778/j.issn.1002-8331.2102-0256.
[4] 卢后洪, 谢罗峰, 朱杨洋, 等. 基于卷积神经网络和模态转换的磁瓦内部缺陷检测方法[J]. 中国测试, 2024, 50(2): 22-27. DOI: 10.11857/j.issn.1674-5124.2022050041.
[5] GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C] //2014 IEEE Conference on Computer Vision and Pattern Recognition. Los Alamitos, CA: IEEE Computer Society, 2014: 580-587. DOI: 10.1109/CVPR.2014.81.
[6] REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6): 1137-1149. DOI: 10.1109/TPAMI.2016.2577031.
[7] 张涛. 基于改进Faster-RCNN的绝缘子缺陷检测模型[J]. 电工技术, 2024(23): 240-243. DOI: 10.19768/j.cnki.dgjs.2024.23.058.
[8] 王子旭, 张红旗, 包曼. 基于Faster-RCNN的绝缘子缺陷检测[J]. 山西电力, 2024(4): 17-21. DOI: 10.3969/j.issn.1671-0320.2024.04.004.
[9] 石炜, 李嘉楠, 张惠丽, 等. 基于Faster R-CNN算法的列车轴承表面缺陷检测研究[J]. 机床与液压, 2021, 49(11): 103-108. DOI: 10.3969/j.issn.1001-3881.2021.11.021.
[10] JIANG P Y, ERGU D J, LIU F Y, et al. A review of yolo algorithm developments[J]. Procedia Computer Science, 2022, 199: 1066-1073. DOI: 10.1016/j.procs.2022.01.135.
[11] QIAN H M, WANG H L, FENG S, et al. FESSD: SSD target detection based on feature fusion and feature enhancement[J]. Journal of Real-Time Image Processing, 2023, 20(1): 2. DOI: 10.1007/s11554-023-01258-y.
[12] 秦振, 李学伟, 刘宏哲. 基于改进SSD的鲁棒小目标检测算法[J]. 东北师大学报(自然科学版), 2023, 55(4):59-66. DOI: 10.16163/j.cnki.dslkxb202209010002.
[13] LI R H, YANG Y, LI N, et al. Transmission line pin detection based on improved SSD[C] //AIIPCC 2022: The Third International Conference on Artificial Intelligence, Information Processing and Cloud Computing. Berlin: VDE Verlag GmbH, 2022: 1-6.
[14] 代培康, 李翰山. 改进YOLOv8n的MEMS缺陷检测方法[J]. 半导体技术, 2025, 50(2): 170-180. DOI: 10.13290/j.cnki.bdtjs.2025.02.009.
[15] 冒浩杰, 巩永旺. 改进YOLOv5s的小目标钢材表面缺陷检测算法[J]. 电子科技, 2025, 38(10): 10-18. DOI:10.16180/j.cnki.issn1007-7820.2025.10.002.
[16] 孙铁强, 刘俊, 宋超, 等. 基于改进YOLOv8n的铝型材表面小目标缺陷检测方法[J]. 现代制造工程, 2024(12): 120-129. DOI: 10.16731/j.cnki.1671-3133.2024.12.016.
[17] 王国国, 龚小玉, 袁发庭. 基于YOLO11s绝缘子缺陷检测的轻量化算法[J/OL]. 重庆工商大学学报(自然科学版): 1-10[2024-12-19]. http://kns.cnki.net/kcms/detail/50.1155.N.20241226.1428.003.html.
[18] SUNKARA R, LUO T. No morestrided convolutions or pooling: a new CNN building block for low-resolution images and small objects[C] //Machine Learning and Knowledge Discovery in Databases: LNAI Volume 13715. Cham: Springer Nature Switzerland, 2023: 443-459. DOI: 10.1007/978-3-031-26409-2_27.
[19] TAN M X, PANG R M, LE Q V.EfficientDet: scalable and efficient object detection[C] //2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). Los Alamitos, CA: IEEE Computer Society, 2020: 10778-10787. DOI: 10.1109/CVPR42600.2020.01079.
[20] LI Y X, LI X, YANG J. Spatial group-wiseenhance: enhancing semantic feature learning in CNN[C] //Computer Vision–ACCV 2022: LNCS Volume 13845. Cham: Springer Nature Switzerland, 2023: 316-332. DOI: 10.1007/978-3-031-26348-4_19.
[21] ZHANG H, XU C, ZHANG S J. Inner-IoU: more effective intersection over union loss with auxiliary bounding box[EB/OL]. (2023-11-14)[2024-12-19]. https://arxiv.org/abs/2311.02877. DOI: 10.48550/arXiv.2311.02877.
[22] KHANAM R, HUSSAIN M. YOLOv11: an overview of the key architectural enhancements[EB/OL].(2024-10-23)[2024-12-19]. https://arxiv.org/abs/2410.17725v1. DOI: 10.48550/arXiv.2410.17725.
[23] 刘作禹. 基于深度学习的高分遥感影像城市道路提取算法研究[D]. 绵阳: 西南科技大学, 2022. DOI: 10.27415/d.cnki.gxngc.2022.001024.
[24] 高民, 陈高华, 古佳欣, 等. FLM-YOLOv8: 一种轻量级的口罩佩戴检测算法[J]. 计算机工程与应用, 2024, 60(17): 203-215. DOI: 10.3778/j.issn.1002-8331.2402-0226.
[25] REZATOFIGHI H, TSOI N, GWAK J, et al. Generalized intersection over union: a metric and a loss for bounding box regression[C] //2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2019: 658-666. DOI: 10.1109/CVPR.2019.00075.
[26] ZHANG Y F, REN W Q, ZHANG Z, et al. Focal and efficient IOU loss for accurate bounding box regression[J]. Neurocomputing, 2022, 506: 146-157. DOI: 10.1016/j.neucom.2022.07.042.
[27] ZHENG Z H, WANG P, LIU W, et al. Distance-IoU loss: faster and better learning for bounding box regression[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(7): 12993-13000. DOI: 10.1609/aaai.v34i07.6999.
[28] GEVORGYAN Z.SIoU loss: more powerful learning for bounding box regression[EB/OL]. (2022-05-25)[2024-12-19]. https://arxiv.org/abs/2205.12740v1. DOI: 10.48550/arXiv.2205.12740.
[29] ZHAO Y A, LV W Y, XU S L, et al. DETRs beat YOLOs on real-time object detection[C] //2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Los Alamitos, CA: IEEE Computer Society, 2024: 16965-16974. DOI: 10.1109/CVPR52733.2024.01605.
[30] ZHAO Y, CHEN B L, LIU B S, et al. GRP-YOLOv5: an improved bearing defect detection algorithm based on YOLOv5[J]. Sensors, 2023, 23(17): 7437. DOI: 10.3390/s23177437.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed