基于改进YOLOv8s轻量化网络的PCBA缺陷检测算法

doi:10.19678/j.issn.1000-3428.0070196

计算机工程 ›› 2026, Vol. 52 ›› Issue (6): 326-338. doi: 10.19678/j.issn.1000-3428.0070196

基于改进YOLOv8s轻量化网络的PCBA缺陷检测算法

沈明辉¹^,², 刘宇杰¹, 陈婧¹, 叶康祈¹, 高赫远¹, 刘建¹^,*(), 姜烨¹, 殷文斐¹, 王恺豪¹, 刘振宇¹

1. 合肥工业大学计算机与信息学院, 安徽宣城 242000
2. 电子科技大学计算机科学与工程学院, 四川成都 611731

收稿日期:2024-08-05 修回日期:2024-10-28 出版日期:2026-06-15 发布日期:2024-12-18
通讯作者: 刘建
作者简介:
沈明辉(CCF学生会员)，男，硕士研究生，主研方向为人工智能算法及其应用
刘宇杰，硕士研究生
陈婧，硕士研究生
叶康祈，硕士研究生
高赫远，学士
刘建(通信作者)，副教授、博士
姜烨，讲师、博士
殷文斐，讲师、博士
王恺豪，学士
刘振宇，学士
基金资助:
国家自然科学基金青年科学基金项目(61906057)

Defect Detection Algorithm for PCBA Based on Improved YOLOv8s Lightweight Network

SHEN Minghui¹^,², LIU Yujie¹, CHEN Jing¹, YE Kangqi¹, GAO Heyuan¹, LIU Jian¹^,*(), JIANG Ye¹, YIN Wenfei¹, WANG Kaihao¹, LIU Zhenyu¹

1. School of Computing and Information Technology, Hefei University of Technology, Xuancheng 242000, Anhui, China
2. School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611731, Sichuan, China

Received:2024-08-05 Revised:2024-10-28 Online:2026-06-15 Published:2024-12-18
Contact: LIU Jian

摘要/Abstract

摘要：

印刷电路板组件(PCBA)的缺陷检测对于提高流水线生产效率具有重要影响, 而在PCBA之后的检查通常由人工完成, 容易造成人力与时间的浪费和出现漏检与误检的情况。为此, 提出一种轻量化的YOLOv8s改进算法, 该算法在有效降低模型复杂度的同时可以提高PCBA缺陷的检测精度。首先, 由于没有公开的PCBA相关数据集, 因此构建一个名为PCBA-DET的组装后PCBA缺陷检测数据集, 并对PCBA-DET进行多种方式的数据增强以模拟实际生产场景和改善数据集的平衡性。其次, 将YOLOv8s的骨干网络最后一个C2f模块替换成重参数化大核卷积网络(RepLKNet)以减少计算开销和提高模型的有效感受野, YOLOv8s的颈部网络引入P2小目标检测层和幽灵卷积以捕捉更多的细节信息, 有效降低模型参数量。实验结果表明, 在数据增强后的PCBA-DET数据集上进行评估, 均值平均精度(mAP)@0.5∶0.95和mAP@0.5与基准模型相比分别上升了2.6和0.1百分点, 但参数量和基线模型相比下降了36.8%。

关键词: 印刷电路板组件, 缺陷检测, YOLOv8算法, Attention-GAN, 轻量化模型

Abstract:

In Printed Circuit Board Assembly (PCBA), defect detection is key to improving production line efficiency. However, after assembly, printed circuit boards are usually inspected manually, leading to labor and time wastage, as well as missed and false detections. To address these issues, this paper proposes an improved lightweight YOLOv8s network that effectively reduces model complexity while enhancing the accuracy of PCBA defect detection. First, owing to the lack of publicly available PCBA-related datasets, a dataset called PCBA-DET is constructed for post-assembly PCBA defect detection. Various data augmentation techniques are applied to PCBA-DET to simulate real-world production scenarios and improve the dataset balance. Second, the last C2f module of the YOLOv8s backbone is replaced with a Re-parameterized Large Kernel convolution Network (RepLKNet) to reduce computational cost and increase the effective receptive field of the model. In addition, in the neck network of YOLOv8s, a P2 small object detection layer and Ghost Convolution are introduced to capture more detailed information and effectively reduce the number of model parameters. On the augmented PCBA-DET dataset, the improved model achieves an increase of 2.6 and 0.1 percentage points in terms of mean Average Precision (mAP)@0.5∶0.95 and mAP@0.5, respectively, compared with the baseline model, whereas the number of parameters is reduced by 36.8%.

Key words: Printed Circuit Board Assembly (PCBA), defect detection, YOLOv8 algorithm, Attention-GAN, lightweight model

沈明辉, 刘宇杰, 陈婧, 叶康祈, 高赫远, 刘建, 姜烨, 殷文斐, 王恺豪, 刘振宇. 基于改进YOLOv8s轻量化网络的PCBA缺陷检测算法[J]. 计算机工程, 2026, 52(6): 326-338.

SHEN Minghui, LIU Yujie, CHEN Jing, YE Kangqi, GAO Heyuan, LIU Jian, JIANG Ye, YIN Wenfei, WANG Kaihao, LIU Zhenyu. Defect Detection Algorithm for PCBA Based on Improved YOLOv8s Lightweight Network[J]. Computer Engineering, 2026, 52(6): 326-338.

https://www.ecice06.com/CN/Y2026/V52/I6/326

图/表 20

图1 PCBA-DET的8种不同的瑕疵种类

Fig.1 Eight different types of defects of PBA-DET

图2 PCBA-DET中的数据增强

Fig.2 Data enhancement in PCBA-DET

图3 PCBA-DET中实例的数量及其所占比例

Fig.3 Number of instances and their share in PCBA-DET

图4 Attention-GAN的网络模型

Fig.4 Network model of Attention-GAN

图5 Attention-GAN生成的多种风扇划痕缺陷

Fig.5 Multiple styles of fan scratch defects generated by Attention-GAN 77.0

图6 替换后的PCBA图

Fig.6 Replaced PCBA diagram

图7 数据增强前后PCBA-DET各项缺陷实例数量变化对比

Fig.7 Comparison of changes in the number of instances of each defect in PCBA-DET before and after data augmentation

图8 改进后的轻量化模型

Fig.8 Improved lightweight model

图9 YOLOv8网络结构

Fig.9 Network structure of YOLOv8 network

图10 YOLOv8s与YOLOv8s-P2颈部网络对比

Fig.10 Comparison of YOLOv8s and YOLOv8s-P2 neck networks

图11 RepLKNet的网络结构

Fig.11 Network structure of RepLKNet

图12 RepLKNet各模块结构

Fig.12 Structure of RepLKNet each modules

图13 传统卷积与Ghost Conv的网络结构对比

Fig.13 Comparison of network structure between conventional convolution and Ghost Conv

图14 改进模型和基线模型对PCBA缺陷检测的结果比较

Fig.14 Comparison of results between improved model and baseline model for PCBA defect detection

参考文献 40

1	ZHANG H X, LI Y F. Integrated optimization of test case selection and sequencing for reliability testing of the mainboard of Internet backbone routers. European Journal of Operational Research, 2022, 299(1): 183- 194. doi: 10.1016/j.ejor.2021.06.028
2	WERNER F T, YILMAZ B B, PRVULOVIC M, et al. Leveraging EM side-channels for recognizing components on a motherboard. IEEE Transactions on Electromagnetic Compatibility, 2020, 63(2): 502- 515. doi: 10.1109/TEMC.2020.3016892
3	CHENG T L, LIU X W, QIN L, et al. A practical micro fringe projection profilometry for 3-D automated optical inspection. IEEE Transactions on Instrumentation and Measurement, 2022, 71, 322- 335.
4	RUNJI J M, LIN C Y. Markerless cooperative augmented reality-based smart manufacturing double-check system: case of safe PCBA inspection following automatic optical inspection. Robotics and Computer-Integrated Manufacturing, 2020, 64, 101957. doi: 10.1016/j.rcim.2020.101957
5	RUNJI J M, LIN C Y. Switchable glass enabled contextualization for a cyber-physical safe and interactive spatial augmented reality PCBA manufacturing inspection system. Sensors, 2020, 20(15): e4286. doi: 10.3390/s20154286
6	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]//Proceedings of the Advances in Neural Information Processing Systems. Cambridge, USA: MIT Press, 2015: 221-232.
7	WU M, AWASTHI N, RAD N M, et al. Advanced ultrasound and photoacoustic imaging in cardiology. Sensors, 2021, 21(23): 7947. doi: 10.3390/s21237947
8	JEON M, YOO S, KIM S W. A contactless PCBA defect detection method: convolutional neural networks with thermographic images. IEEE Transactions on Components, Packaging and Manufacturing Technology, 2022, 12(3): 489- 501. doi: 10.1109/TCPMT.2022.3147319
9	GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Columbus, USA: IEEE Press, 2014: 580-587.
10	GIRSHICK R. Fast R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2015: 1440-1448.
11	REN S, HE K, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[C]//Proceedings of the Advances in Neural Information Processing Systems. Cambridge, USA: MIT Press, 2015: 28-37.
12	HE K M, GKIOXARI G, DOLLAR P, et al. Mask R-CNN[C]//Proceedings of the IEEE International Conference on Computer Vision. Washington D. C., USA: IEEE Press, 2017: 2961-2969.
13	DAI J, LI Y, HE K, et al. R-FCN: object detection via region-based fully convolutional networks[C]//Proceedings of the Advances in Neural Information Processing Systems. Cambridge, USA: MIT Press, 2016: 29.
14	CAI Z, VASCONCELOS N. Cascade R-CNN: high quality object detection and instance segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(5): 1483- 1498. doi: 10.1109/TPAMI.2019.2956516
15	ZHANG H Y, WANG Y, DAYOUB F, et al. VarifocalNet: an IoU-aware dense object detector[C]// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville, USA: IEEE Press, 2021: 8514-8523.
16	LUO J X, YANG Z Y, LI S P, et al. FPCB surface defect detection: a decoupled two-stage object detection framework. IEEE Transactions on Instrumentation and Measurement, 2021, 70, 432- 441. doi: 10.1109/TIM.2021.3092510
17	LIU W, ANGUELOV D, ERHAN D, et al. SSD: single shot MultiBox detector[M]. Berlin, Germany: Springer, 2016.
18	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. Las Vegas, USA: IEEE Press, 2016: 779-788.
19	VARGHESE R, SAMBATH M. YOLOv8: a novel object detection algorithm with enhanced performance and robustness[C]// Proceedings of the International Conference on Advances in Data Engineering and Intelligent Computing Systems. Washington D. C., USA: IEEE Press, 2024: 531-543.
20	SHEN M, LIU Y, CHEN J, et al. Defect detection of printed circuit board assembly based on YOLOv5. Scientific Reports, 2024, 14(1): 19287. doi: 10.1038/s41598-024-70176-1
21	BHATTACHARYA A, CLOUTIER S G. End-to-end deep learning framework for printed circuit board manufacturing defect classification. Scientific Reports, 2022, 12(1): 12559. doi: 10.1038/s41598-022-16302-3
22	WANG C Y, BOCHKOVSKIY A, LIAO H M. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington D. C., USA: IEEE Press, 2023: 7464-7475.
23	PARK J H, KIM Y S, SEO H, et al. Analysis of training deep learning models for PCB defect detection. Sensors, 2023, 23(5): 2766. doi: 10.3390/s23052766
24	赵小虎, 谢礼逊, 慕灯聪, 等. 基于TCM-YOLO网络的金属表面缺陷检测方法. 计算机工程, 2025, 51(6): 338- 348. doi: 10.19678/j.issn.1000-3428.0069285
	ZHAO X H, XIE L X, MU D C, et al. Metal surfacedefect detection based on TCM-YOLO network methods. Computer Engineering, 2025, 51(6): 338- 348. doi: 10.19678/j.issn.1000-3428.0069285
25	LEMLEY J, BAZRAFKAN S, CORCORAN P. Smart augmentation learning an optimal data augmentation strategy. IEEE Access, 2017, 5, 5858- 5869. doi: 10.1109/ACCESS.2017.2696121
26	秦涛, 王云龙, 孙哲南, 等. 基于深度学习的眼周识别方法研究. 计算机应用与软件, 2024, 41(6): 186- 193. doi: 10.3969/j.issn.1000-386x.2024.06.028
	QIN T, WANG Y L, SUN Z N, et al. Research on eye perimeter recognition method based on deep learning. Computer Applications and Software, 2024, 41(6): 186- 193. doi: 10.3969/j.issn.1000-386x.2024.06.028
27	王帅, 史艳翠. 基于个性化数据增强的自监督序列推荐算法. 计算机工程, 2025, 51(8): 190- 202. doi: 10.19678/j.issn.1000-3428.0069636
	WANG S, SHI Y C. Personalized data augmentation for self-supervised sequential recommendation. Computer Engineering, 2025, 51(8): 190- 202. doi: 10.19678/j.issn.1000-3428.0069636
28	KIM M, JEONG C Y. Label-preserving data augmentation for mobile sensor data. Multidimensional Systems and Signal Processing, 2021, 32(1): 115- 129. doi: 10.1007/s11045-020-00731-2
29	TRAN N T, TRAN V H, NGUYEN N B, et al. On data augmentation for GAN training. IEEE Transactions on Image Processing, 2021, 30, 1882- 1897. doi: 10.1109/TIP.2021.3049346
30	DU Z W, GAO L, LI X Y. A new contrastive GAN with data augmentation for surface defect recognition under limited data. IEEE Transactions on Instrumentation and Measurement, 2023, 72, 1- 13.
31	PU Z Q, CABRERA D, LI C, et al. Sliced wasserstein cycle consistency generative adversarial networks for fault data augmentation of an industrial robot. Expert Systems with Applications, 2023, 222, 119754. doi: 10.1016/j.eswa.2023.119754
32	CAO J, LUO M, YU J, et al. ScoreMix: a scalable augmentation strategy for training GANs with limited data. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(7): 8920- 8935. doi: 10.1109/TPAMI.2022.3231649
33	EMAMI H, ALIABADI M M, DONG M, et al. SPA-GAN: spatial attention GAN for image-to-image translation. IEEE Transactions on Multimedia, 2021, 23, 391- 401. doi: 10.1109/TMM.2020.2975961
34	CHEN X Y, XU C, YANG X K, et al. Attention-GAN for object transfiguration in wild images[C]//Proceedings of the European Conference on Computer Vision. Berlin, Germany: Springer, 2018: 167-184.
35	JANSI RANI S V, IOANNOU I, SWETHA R, et al. A novel automated approach for fish biomass estimation in turbid environments through deep learning, object detection, and regression. Ecological Informatics, 2024, 81, 102663. doi: 10.1016/j.ecoinf.2024.102663
36	贵向泉, 刘世清, 李立, 等. 基于改进YOLOv8的景区行人检测算法. 计算机工程, 2024, 50(7): 342- 351. doi: 10.19678/j.issn.1000-3428.0068125
	GUI X Q, LIU S Q, LI L, et al. Pedestrian detection algorithm for scenic spots based on improved YOLOv8. Computer Engineering, 2024, 50(7): 342- 351. doi: 10.19678/j.issn.1000-3428.0068125
37	CHOLLET F. Xception: deep learning with depthwise separable convolutions[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, USA: IEEE Press, 2017: 1251-1258.
38	DING X H, ZHANG X Y, HAN J G, et al. Scaling up your kernels to 31×31: revisiting large kernel design in CNNs[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans, USA: IEEE Press, 2022: 11963-11975.
39	HAN K, WANG Y H, TIAN Q, et al. GhostNet: more features from cheap operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle, USA: IEEE Press, 2020: 1580-1589.
40	ZHANG Z Z, LI J W, SU C W, et al. A method for counting fish based on improved YOLOv8. Aquacultural Engineering, 2024, 107, 102450.

[1]	唐克, 魏飞鸣, 李东瀛, 郁文贤. 基于改进YOLOv8的轻量化无人机图像目标检测算法[J]. 计算机工程, 2026, 52(3): 97-106.
[2]	马跃, 黄周睿, 周雯, 许艺瀚. 基于感受野注意力的轻量化林火检测算法[J]. 计算机工程, 2025, 51(9): 350-361.
[3]	倪源松, 韩军, 邹小燕, 胡广怡, 王文帅. 两阶段自适应分块输电线路螺栓缺陷检测方法[J]. 计算机工程, 2025, 51(8): 281-291.
[4]	彭菊红, 张弛, 高谦, 张光明, 谈栋华, 赵明俊. 基于改进的YOLOv8算法的钢材缺陷检测[J]. 计算机工程, 2025, 51(7): 152-160.
[5]	张佳承, 韦锦, 陈义时. 改进YOLOv8的实时轻量化鲁棒绿篱检测算法[J]. 计算机工程, 2025, 51(7): 362-374.
[6]	赵小虎, 谢礼逊, 慕灯聪, 张悦. 基于TCM-YOLO网络的金属表面缺陷检测方法[J]. 计算机工程, 2025, 51(6): 338-348.
[7]	黄硕清, 黄金贵. 基于RFB和YOLOv5特征增强融合改进的钢材表面缺陷检测方法[J]. 计算机工程, 2025, 51(4): 249-260.
[8]	汤静雯, 赖惠成, 王同官. 远距离情形下的改进YOLOv8行人检测算法[J]. 计算机工程, 2025, 51(4): 303-313.
[9]	王奕丰, 曾诚, 全擎宇, 王娇然, 何鹏. 基于多特征融合的智能合约缺陷检测方法[J]. 计算机工程, 2024, 50(8): 133-141.
[10]	贵向泉, 刘世清, 李立, 秦庆松, 李唐艳. 基于改进YOLOv8的景区行人检测算法[J]. 计算机工程, 2024, 50(7): 342-351.
[11]	张诗婧, 莫绪涛, 赵行, 董杨林. 基于球面折反射成像和YOLOv7的内螺纹缺陷检测[J]. 计算机工程, 2024, 50(7): 282-292.
[12]	赵继达, 甄国涌, 储成群. 基于YOLOv8的无人机图像目标检测算法[J]. 计算机工程, 2024, 50(4): 113-120.
[13]	白宇, 王珺, 冉红雷, 安胜彪. 半导体器件内部缺陷标注与检测方法研究[J]. 计算机工程, 2024, 50(12): 245-253.
[14]	兰红, 王惠钊. 结合轻量化与多尺度融合的交通标志检测算法[J]. 计算机工程, 2024, 50(10): 381-392.
[15]	季娟娟, 王佳, 陈亚杰, 卢道华. 基于改进YOLOv7的液压阀块表面微小缺陷检测[J]. 计算机工程, 2023, 49(11): 302-310.

选择文件类型/文献管理软件名称

选择包含的内容

基于改进YOLOv8s轻量化网络的PCBA缺陷检测算法

Defect Detection Algorithm for PCBA Based on Improved YOLOv8s Lightweight Network

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 20

参考文献 40

相关文章 15

编辑推荐

Metrics

本文评价

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

基于改进YOLOv8s轻量化网络的PCBA缺陷检测算法

Defect Detection Algorithm for PCBA Based on Improved YOLOv8s Lightweight Network

RichHTML

PDF

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 20

参考文献 40

相关文章 15

编辑推荐

Metrics

本文评价