YOLOv11-MAS：一种高效的PCB缺陷检测算法

doi:10.3778/j.issn.1002-8331.2503-0054

摘要/Abstract

摘要： 随着电子设备对印刷电路板（printed circuit board，PCB）的要求越来越高，在PCB加工过程中的缺陷检测工作也愈发重要，但由于缺陷特征目标小、类别复杂且难以区分，传统检测方法在精度和鲁棒性方面存在局限。为此，对最新的YOLOv11版本改进，形成了PCB缺陷检测模型——YOLOv11-MAS。在骨干网络部分设计并嵌入中值增强空间通道注意力模块（median-enhanced channel and spatial attention block，MECS），以中值增强与深度卷积的方式扩大感受野，提高对PCB电路板特殊缺陷的识别能力。在颈部网络部分，采用自适应分层特征融合网络（adaptive hierarchical feature integration network，AHFIN），通过自适应加权融合多尺度特征，以增强对主干网络信息的利用，使模型能够精准关注关键缺陷区域。此外，将YOLOv11的损失函数替换为滑动对齐损失函数（slide alignment loss，SAL），通过优化边界框的界定，提高对复杂缺陷类型的检测能力。实验结果表明，YOLOv11-MAS模型在PCB缺陷数据集中的多个指标均表现优异，mAP值达到93.1%，较YOLOv11提升了7.8个百分点，同时在NVIDIA Jetson Xavier嵌入式设备上的检测帧率达49.6帧/s，能满足工业实时检测需求。

关键词: 缺陷检测, 小目标检测, YOLOv11, 注意力机制, 损失函数

Abstract: PCB defect detection is crucial in electronic engineering. However, traditional detection methods face limitations in accuracy and robustness due to the small size, complex categories, and indistinguishability of defect features. To address these challenges, this paper proposes an improved YOLOv11-based PCB defect detection model: YOLOv11-MAS. The model incorporates a median-enhanced channel and spatial attention block (MECS) in the backbone to enhance defect feature extraction through an improved attention mechanism. Median enhancement and depth wise convolution are employed to expand the receptive field and improve the recognition of various special defects on PCBs. In the neck part, an adaptive hierarchical feature integration network (AHFIN) is introduced to adaptively weight and fuse multi-scale features, thereby enhancing the utilization of information from the backbone and enabling the model to focus precisely on key defect areas. Additionally, the CIoU loss function of YOLOv11 is replaced with a slide alignment loss (SAL) function. By optimizing bounding box definition and considering center deviation, size ratio, and angle differences, SAL introduces adaptive weights for difficult and easy samples to accurately regress target boxes and improve detection capabilities for complex defect types. Experimental results demonstrate that the YOLOv11-MAS model achieves excellent performance on multiple PCB defect datasets, with an mAP of 93.1%, a 7.8 percentage points improvement over YOLOv11, and a detection speed of 49. 6 frames per second on the NVIDIA Jetson Xavier, which meets the requirements for real-time detection.

Key words: defect detection, small target detection, YOLOv11, attention mechanism, loss function

殷旭鹏, 赵兴强. YOLOv11-MAS：一种高效的PCB缺陷检测算法[J]. 计算机工程与应用, 2025, 61(17): 102-111.

YIN Xupeng, ZHAO Xingqiang. YOLOv11-MAS: Efficient PCB Defect Detection Algorithm[J]. Computer Engineering and Applications, 2025, 61(17): 102-111.

参考文献

[1] CHEN I C, HWANG R C, HUANG H C. PCB defect detection based on deep learning algorithm[J]. Processes, 2023, 11(3): 775.
[2] DENG L, MENG X Z, GUO X D, et al. Learned optical flow-based defect detection with coarse-to-fine nonrigid image alignment[J]. IEEE Sensors Journal, 2025, 25(5): 8684-8696.
[3] FELICIANO F, LETA F, MARTINS F. Computer vision system for printed circuit board inspection[J]. ABCM Symposium Series in Mechatronics, 2018, 3: 623-632.
[4] KAMALAJA N, SILVA D, TANTHIRIGE M. Printed circuit board defect detection and sorting using image processing techniques[J]. International Journal of Engineering Research in Electronic and Communication Engineering, 2016, 3(2): 78-90.
[5] ZHOU Y B, YUAN M H, ZHANG J, et al. Review of vision-based defect detection research and its perspectives for printed circuit board[J]. Journal of Manufacturing Systems, 2023, 70: 557-578.
[6] SILVA L H, AZEVEDO G O, FERNANDES B J T, et al. Automatic optical inspection for defective PCB detection using transfer learning[C]//Proceedings of the IEEE Latin American Conference on Computational Intelligence. Piscataway: IEEE, 2019: 1-6.
[7] WAN Y, GAO L, LI X, et al. Semi-supervised defect detection method with data-expanding strategy for PCB quality inspection[J]. Sensors (Basel), 2022, 22(20): 7971.
[8] VIRASOVA A Y, KLIMOV D I, KHROMOV O E, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[J]. Radioengineering, 2021: 115-126.
[9] 颜豪男, 吕伏. 多尺度特征增强的PCB板表面缺陷检测算法[J/OL]. 信息与控制: 1-13[2025-08-20]. https://doi.org/10. 13976/j.cnki.xk.2024.4021.
YAN H N, LYU F. Multi-scale feature enhancement for PCB defect detection algorithm[J]. Information and Control: 1-13[2025-08-20]. https://doi.org/10.13976/j.cnki.xk.2024.4021.
[10] LONDE P, CHAVAN S. Automatic PCB defects detection and classification using matlab[J]. International Journal of Current Engineering and Technology, 2014, 4: 78-90.
[11] 向毅伟, 蒋瑜, 王琪凯, 等. 多尺度特征优化的实时Transformer在无人机航拍中的研究[J]. 计算机工程与应用, 2025, 61(9): 221-229.
XIANG Y W, JIANG Y, WANG Q K, et al. Research on real-time transformer for multi-scale feature optimization in drone aerial imaging[J]. Computer Engineering and Applications, 2025, 61(9): 221-229.
[12] ZHANG K H, SHEN H K. Solder joint defect detection in the connectors using improved faster-RCNN algorithm[J]. Applied Sciences, 2021, 11(2): 576.
[13] 吴一全, 赵朗月, 苑玉彬, 等. 基于机器视觉的PCB缺陷检测算法研究现状及展望[J]. 仪器仪表学报, 2022, 43(8): 1-17.
WU Y Q, ZHAO L Y, YUAN Y B, et al. Research status and the prospect of PCB defect detection algorithm based on machine vision[J]. Chinese Journal of Scientific Instrument, 2022, 43(8): 1-17.
[14] RAN G Z, LEI X, LI D S, et al. Research on PCB defect detection using deep convolutional nerual network[C]//Proceedings of the 5th International Conference on Mechanical, Control and Computer Engineering. Piscataway: IEEE, 2020: 1310-1314.
[15] ANITHA D B, RAO M. A survey on defect detection in bare PCB and assembled PCB using image processing techniques[C]//Proceedings of the International Conference on Wireless Communications, Signal Processing and Networking. Piscataway: IEEE, 2017: 39-43.
[16] LING Q, ISA N A M. Printed circuit board defect detection methods based on image processing, machine learning and deep learning: a survey[J]. IEEE Access, 2023, 11: 15921-15944.
[17] ZHANG H, JIANG L X, LI C Q. CS-ResNet: cost-sensitive residual convolutional neural network for PCB cosmetic defect detection[J]. Expert Systems with Applications, 2021, 185: 115673.
[18] 潘玮, 韦超, 钱春雨, 等. 面向无人机视角下小目标检测的YOLOv8s改进模型[J]. 计算机工程与应用, 2024, 60(9): 142-150.
PAN W, WEI C, QIAN C Y, et al. Improved YOLOv8s model for small object detection from perspective of drones[J]. Computer Engineering and Applications, 2024, 60(9): 142-150.
[19] 窦智, 高浩然, 刘国奇, 等. 轻量化YOLOv8的小样本钢板缺陷检测算法[J]. 计算机工程与应用, 2024, 60(9): 90-100.
DOU Z, GAO H R, LIU G Q, et al. Small sample steel plate defect detection algorithm of lightweight YOLOv8[J]. Computer Engineering and Applications, 2024, 60(9): 90-100.
[20] 胡峻峰, 李柏聪, 朱昊, 等. 改进YOLOv8的轻量化无人机目标检测算法[J]. 计算机工程与应用, 2024, 60(8): 182-191.
HU J F, LI B C, ZHU H, et al. Improved YOLOv8 lightweight UAV target detection algorithm[J]. Computer Engineering and Applications, 2024, 60(8): 182-191.
[21] REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: unified, real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 779-788.
[22] ADIBHATLA V A, CHIH H C, HSU C C, et al. Applying deep learning to defect detection in printed circuit boards via a newest model of you-only-look-once[J]. Mathematical Biosciences and Engineering, 2021, 18(4): 4411-4428.
[23] 赵奕翔, 苏检德, 何富运, 等. 改进YOLOv5s算法的PCB缺陷检测方法[J]. 无线电工程, 2025, 55(5): 949-958.
ZHAO Y X, SU J D, HE F Y, et al. PCB defect detection method based on improved YOLOv5s algorithm[J]. Radio Engineering, 2025, 55(5): 949-958.
[24] 解琳, 韩跃平, 翟倩, 等. 基于改进YOLOv7-tiny的PCB表面缺陷检测[J]. 测试技术学报, 2025, 39(1): 81-87.
XIE L, HAN Y P, ZHAI Q, et al. PCB surface defect detection based on improved YOLOv7-tiny[J]. Journal of Testing Technology, 2025, 39(1): 81-87.
[25] 吕秀丽, 杨昕升, 曹志民. 改进YOLOv8的PCB表面缺陷检测算法[J]. 电子测量技术, 2024, 47(12): 100-108.
LYU X L, YANG X S, CAO Z M. PCB surface defect dete-ction algorithm based on improved YOLOv8[J]. Electronic Measurement Technology, 2024, 47(12): 100-108.