目的 解决轴承表面缺陷检测中目标尺寸小、背景复杂、检测精度与速度难以平衡的技术难题。方法 基于YOLOv8n框架构建改进而来的轻量化检测算法,通过骨干网络中以GSConv模块替换C2F结构实现模型压缩,引入无需额外参数的SimAM注意力机制,有效提升缺陷特征提取能力。采用渐进式特征金字塔网络（AFPN）优化多尺度特征融合效率,将边界框回归损失函数改进为MPDIoU,以提升小目标定位精度。结果 改进后的模型在轴承缺陷数据集上取得92.32%的mAP@0.5检测精度,检测速度提升至121.8 FPS,模型参数量较原始YOLOv8n减少16.25%。通过消融实验验证各改进模块的有效性,相较于原始YOLOv8n模型,改进后模型在保持实时性的同时,实现了12.5%的Precision有效提升。结论 构建的轻量化改进算法有效平衡了检测精度与速度矛盾,参数量减少带来的计算效率提升未影响检测性能,改进后的综合指标满足工业现场对轴承表面缺陷高精度实时检测的需求,为解决小目标工业缺陷检测问题提供了新的技术方案。
Abstract
The work aims to address the technical challenges of small defect sizes, complex backgrounds, and the accuracy-speed trade-off in bearing surface defect detection. Based on the lightweight detection algorithm improved based on the YOLOv8n framework, the backbone’s C2F structure was replaced with a Group Spatial-Shuffle Convolution (GSConv) module to reduce model complexity and parameters. A parameter-free SimAM attention mechanism was integrated to enhance defect feature localization. An Asymptotic Feature Pyramid Network (AFPN) was applied to optimize multi-scale feature fusion efficiency. The bounding box regression loss was improved to Minimum Point Distance Intersection over Union (MPDIoU) for precise small-target positioning. Experimental results demonstrated a mAP@0.5 of 92.32% at 121.8 FPS on bearing defect datasets. The number of model parameters was 16.25% less than that of the original YOLOv8n. Ablation studies validated the contributions of each module. Compared with the original YOLOv8n model, the improved model achieved a 12.5% accuracy improvement over industrial inspection benchmarks while maintaining real-time performance. The proposed method effectively balances high-precision detection and computational efficiency, meets the requirements for real-time industrial defect inspection, and provides a novel technical solution for small-target defect identification in manufacturing quality control.
关键词
缺陷检测 /
YOLOv8 /
轴承表面缺陷检测 /
注意力机制 /
小目标检测 /
轻量级网络
Key words
defect detection /
YOLOv8 /
bearing surface defect detection /
attention mechanism /
small-target detection /
lightweight network
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] MA J J, LIU M L, HU S Y, et al.A Novel CNN Ensemble Framework for Bearing Surface Defects Classification Based on Transfer Learning[J]. Measurement Science and Technology, 2023, 34(2): 025902.
[2] KUMAR P, KUMAR P, HATI A S, et al.Deep Transfer Learning Framework for Bearing Fault Detection in Motors[J]. Mathematics, 2022, 10(24): 4683.
[3] 张寅, 朱桂熠, 施天俊, 等. 基于特征融合与注意力的遥感图像小目标检测[J]. 光学学报, 2022, 42(24): 2415001.
ZHANG Y, ZHU G Y, SHI T J, et al.Small Object Detection in Remote Sensing Images Based on Feature Fusion and Attention[J]. Acta Optica Sinica, 2022, 42(24): 2415001.
[4] REDMON J, DIVVALA S, GIRSHICK R, et al.You Only Look Once: Unified, Real-Time Object Detection[C]// Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE, 2016.
[5] REDMON J, FARHADI A.YOLO9000: Better, Faster, Stronger[C]// Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu: IEEE, 2017.
[6] REDMON J, FARHADI A. YOLOv3: An Incremental Improvement[EB/OL]. (2018-04-08)[2025-05-21]. https://arxiv.org/abs/1804.02767.
[7] BOCHKOVSKIY A, WANG C Y, LIAO H Y M. YOLOv4: Optimal Speed and Accuracy of Object Detection[EB/OL]. (2020-04-23)[2025-05-21]. https://arxiv.org/abs/2004.10934.
[8] GE Z, LIU S, WANG F, et al. YOLOX: Exceeding YOLO Series in2021[EB/OL]. (2021-07-18)[2025-05-21]. https://arxiv.org/abs/2107.08430.
[9] LI C, LI L, JIANG H, et al. YOLOv6: A Single-Stage Object Detection Framework for Industrial Applications[EB/OL]. (2022-09-07)[2025-05-21]. https://arxiv.org/abs/2209.02976.
[10] WANG C Y, BOCHKOVSKIY A, LIAO H Y M. YOLOv7: Trainable bag-of-freebies sets new state-of-the- art for real-time object detectors[EB/OL]. (2022-07-06) [2025-05-21]. https://arxiv.org/abs/2207.02696.
[11] LIU W, ANGUELOV D, ERHAN D, et al.SSD: Single Shot MultiBox Detector[C]// Proceedings of Computer Vision—ECCV 2016. Cham: Springer International Publishing, 2016.
[12] LIN T Y, GOYAL P, GIRSHICK R, et al.Focal Loss for Dense Object Detection[C]// Proceedings of 2017 IEEE International Conference on Computer Vision (ICCV). Venice: IEEE, 2017.
[13] GIRSHICK R.Fast R-CNN[C]// Proceedings of 2015 IEEE International Conference on Computer Vision (ICCV). Santiago: IEEE, 2015.
[14] REN S Q, HE K M, GIRSHICK R, et al.Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks[C]// Proceedings of the 28th International Conference on Neural Information Processing Systems. [s. l.]: IEEE, 2017.
[15] HE K M, GKIOXARI G, DOLLÁR P, et al. Mask R- CNN[C]// Proceedings of 2017 IEEE International Conference on Computer Vision (ICCV). Venice: IEEE, 2017.
[16] CARION N, MASSA F, SYNNAEVE G, et al.End-to- End Object Detection with Transformers[M]. Cham: Springer International Publishing, 2020: 213-229.
[17] ZHU X, SU W, LU L, et al.Deformable DETR: Deformable Transformers for End-to-End Object Detection[J]. arXiv:2010.04159, 2020.
[18] 吴凤和, 崔健新, 张宁, 等. 基于改进YOLOv4算法的轮毂表面缺陷检测[J]. 计量学报, 2022, 43(11): 1404-1411.
WU F H, CUI J X, ZHANG N, et al.Surface Defect Detection of Wheel Hub Based on Improved YOLOv4 Algorithm[J]. Acta Metrologica Sinica, 2022, 43(11): 1404-1411.
[19] 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.
YAO J L, CHENG G, WAN F, et al.Improved Lightweight Bearing Defect Detection Algorithm of YOLOv8[J]. Computer Engineering and Applications, 2024, 60(21): 205-214.
[20] 张利丰, 田莹. 改进YOLOv8的多尺度轻量型车辆目标检测算法[J]. 计算机工程与应用, 2024, 60(3): 129-137.
ZHANG L F, TIAN Y.Improved YOLOv8 Multi-Scale and Lightweight Vehicle Object Detection Algorithm[J]. Computer Engineering and Applications, 2024, 60(3): 129-137.
[21] LI H, LI L, WEI H, et al.Slim-Neck by GsConv: A Better Design Paradigm of Detector Architectures for Autonomous Vehicles[J]. Journal of Real-Time Image Processing, 2024, 21: 62.
[22] YANG L, ZHANG R Y, LI L, et al.SimAM: A Simple, Parameter-Free Attention Module for Convolutional Neural Networks[C]// Proceedings of the 38th International Conference on Machine Learning. [s. l.]: PMLR, 2021.
[23] YANG G Y, LEI J, ZHU Z K, et al.AFPN: Asymptotic Feature Pyramid Network for Object Detection[C]// 2023 IEEE International Conference on Systems, Man, and Cybernetics (SMC). Honolulu: IEEE, 2023.
[24] 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.
[25] ELFWING S, UCHIBE E, DOYA K.Sigmoid-Weighted Linear Units for Neural Network Function Approximation in Reinforcement Learning[J]. Neural Networks, 2018, 107: 3-11.
[26] CHENG Q Q, LI X H, ZHU B, et al.Drone Detection Method Based on MobileViT and CA-PANet[J]. Electronics, 2023, 12(1): 223.
[27] WAN Q, HUANG Z L, LU J C, et al. SeaFormer: Squeeze-Enhanced Axial Transformer for Mobile Semantic Segmentation[EB/OL]. (2023-01-30)[2025-05-21]. https://arxiv.org/abs/2301.13156.
[28] 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.
[29] GEVORGYAN Z. SIoU Loss: More Powerful Learning for Bounding Box Regression[EB/OL]. (2022-05-25) [2025-05-21]. https://arxiv.org/abs/2205.12740.
[30] REZATOFIGHI H, TSOI N, GWAK J, et al.Generalized Intersection over Union: A Metric and a Loss for Bounding Box Regression[C]// Proceedings of 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach: IEEE, 2019.
PDF(2039 KB)
