重庆科技大学学报（自然科学版）2025，Vol.27Issue(5)：91-101,11.DOI:10.19406/j.issn.2097-4531.2025.05.011
基于DBC-YOLOv8s改进模型的PCB缺陷检测方法研究
Research on PCB Defect Detection Method Based on Improved DBC-YOLOv8s Model
摘要
Abstract
In the quality inspection process of printed circuit boards,the precise identification of minute target de-fects faces numerous challenges,primarily manifested in extremely small object sizes and low contrast between de-fect features and background areas.To address these issues,an improved object detection model,DBC-YOLOv8s,is proposed.Firstly,the C2f_DCNv2 module is created by integrating DCNv2 with C2f in the feature fusion layer of the backbone network,enabling precise capture of defect features in irregular and complex backgrounds.Secondly,the BiFPN is introduced to enhance the network's fusion efficiency for multi-scale features.Finally,the CBAM is embedded to focus on target regions,improving the model's detection accuracy and robustness.Experimental re-sults demonstrate that this model achieves improved detection accuracy and speed while maintaining lightweight characteristics.关键词
印刷电路板/表面缺陷识别/YOLOv8s算法/目标检测Key words
printed circuit board/surface defect recognition/YOLOv8s algorithm/object detection分类
信息技术与安全科学引用本文复制引用
陈松,周婷,陈立爱,张谦..基于DBC-YOLOv8s改进模型的PCB缺陷检测方法研究[J].重庆科技大学学报（自然科学版）,2025,27(5):91-101,11.基金项目
安徽省高等学校科学研究重大项目"基于稀疏自编码器和卷积神经网络的输送机复合故障诊断方法及应用研究"(2022AH040044) （2022AH040044）
