YOLO-PMAA基于YOLOv11的增强PCB小目标缺陷检测算法
(湖南工程职业技术学院,湖南 长沙 410151)
摘要
随着PCB朝着高密度、复杂化、小型化发展,对复杂背景下小目标缺陷检测提出高要求。为了提升模型小目标缺陷检出能力,基于YOLOv11模型,提出了一种增强PCB小目标检测算法YOLO-PMAA。首先,在骨干网络浅层引入风车型卷积PConv,通过非对称填充与卷积扩大感受野,确保小目标特征向网络深层传递;其次,设计了多维级联混合注意力模块MDCA,通过尺度、通道、空间多维度关注重要信息,加强对细、长、窄类型小目标的特征提取;最后,设计了自适应形状损失函数ASIoU,提升模型检测性能。该算法在PKU-Market-PCB数据集上进行多轮验证,与YOLOv11n相比,精确率提升3.0%,召回率提升6.2%,mAP@0.5提升3.8%,mAP@0.5:0.95提升3.2%,参数量减少4.25%,权重文件下降3.82%,在平衡模型复杂度的同时,有效提升了模型性能。
关键词
PCB缺陷检测;注意力机制;损失函数;特征融合;小目标检测
参考文献
- [1] 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.
- [2] Xia K W,Lyu Z L,Liu K,et al. Global Contextual Attention Augmented YOLO with ConvMixer Prediction Heads for PCB Surface Defect Detection [J].Scientific Reports,2023,13(1):9805-9805.
- [3] Li B,Palayew S,Li F,et al. PCBDet: An Efficient Deep Neural Network Object Detection Architecture for Automatic PCB Component Detection on the Edge [J/OL].Electronics Letters,2024,60(2):e13042.https://doi.org/10.1049/ell2.13042.
- [4] LI D,BAI F,LI S,et al. EL-PCBNet: An Efficient and Lightweight Network for PCB Defect Detection [J/OL].Measurement,2025,253:11771910.1016/j.measurement.2025.117719.
- [5] Zheng H,Peng J,Yu X,et al. FDDC-YOLO: an efficient detection algorithm for dense small-target solder joint defects in PCB inspection [J/OL].Journal of Real-Time Image Processing,2025,22(2):83.https://link.springer.com/article/10.1007/s11554-025-01664-4.
- [6] Xiao G S,Hou S L,Zhou H Y. PCB Defect Detection Algorithm Based on CDI-YOLO [J/OL].Scientific Reports,2024,14(1):https://doi.org/10.1038/s41598-024-57491-3.
- [7] 魏嘉敏,续欣莹,韩晖,等.基于YOLO-HorL的轻量级PCB缺陷检测算法 [J].计算机工程与设计,2025,46(11):3283-3291.
- [8] Yang J N,Liu S L,Wu J J,et al. Pinwheel-shaped Convolution and Scale-based Dynamic Loss for Infrared Small Target Detection [C]//Proceedings of the AAAI Conference on Artificial Intelligence,2025,39(9):9202-9210.
- [9] Zhang Y,Zhang H,Zhao H M. DsP-YOLO: An anchor-free network with DsPAN for small object detection of multiscale defects [J/OL].Expert Systems with Applications,2024,241:122669.https://doi.org/10.1016/j.eswa.2023.122669.
- [10] Zhu H Y,Wei L Y,Li H. VR-YOLO: Enhancing PCB Defect Detection with Viewpoint Robustness Based on YOLO [C]//2025 IEEE 11th World Forum on Internet of Things (WF-IoT),2025:1-6.
基金项目
湖南省教育厅科学研究项目(25C0994)
作者简介
周杨淼(1991—),女,回族,湖北武汉人,讲师,硕士,研究方向:计算机视觉。