Article(id=1319955554672431797, tenantId=1146029695717560320, journalId=1278651732997652489, issueId=1319955375944745964, articleNumber=null, orderNo=null, doi=10.12086/oee.2026.250117, pmid=null, cstr=32245.14.oee.2026.250117, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1744992000000, receivedDateStr=2025-04-19, revisedDate=1750608000000, revisedDateStr=2025-06-23, acceptedDate=1750608000000, acceptedDateStr=2025-06-23, onlineDate=1792631527821, onlineDateStr=2026-10-22, pubDate=1771948800000, pubDateStr=2026-02-25, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1792631527821, onlineIssueDateStr=2026-10-22, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1792631527821, creator=13701087609, updateTime=1792631527821, updator=13701087609, issue=Issue{id=1319955375944745964, tenantId=1146029695717560320, journalId=1278651732997652489, year='2026', volume='53', issue='2', pageStart='250117', pageEnd='1', issueExtLink='null', onlineDate='null', pubDate='1771948800000', pubDateStr='2026-02-25', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1792631485210, creator='13701087609', updateTime=1792631485210, updator='13701087609', preIssue=null, nextIssue=null, articleTotal=null, ext=null, issueFiles=null, downloadFileDto=null}, startPage=250117, endPage=1, ext={EN=ArticleExt(id=1319955554869564086, articleId=1319955554672431797, tenantId=1146029695717560320, journalId=1278651732997652489, language=EN, title=Printed circuit board defect detection based on LightRT-DETR algorithm, columnId=1279511634116841602, journalTitle=Opto-Electronic Engineering, columnName=Article, runingTitle=null, highlight=null, articleAbstract=
Objective

The manufacturing of printed circuit boards (PCBs) constitutes the foundational infrastructure of the contemporary electronics industry. As the sector evolves in response to consumer demands for increasingly miniaturized, sophisticated, and high-performance devices, the physical characteristics of PCBs have undergone significant transformation. Modern boards are characterized by high-density interconnects, multi-layered architectures, and microscopic component placement. In this context, the role of quality control has shifted from a supplementary process to a critical determinant of product viability and yield. Traditionally, defect detection has relied on manual inspection or standard automated optical inspection (AOI) systems. However, these conventional methods are increasingly challenged by the escalating complexity of modern PCB designs. Manual inspection is labor-intensive, subjective, and prone to fatigue-induced errors, rendering it unsuitable for mass production. Similarly, traditional AOI systems based on pattern matching or simple image processing often struggle with the subtle variations and complex backgrounds found in high-density PCBs. Consequently, there is a growing necessity for more intelligent and adaptive inspection solutions capable of discerning defects with high precision. Current state-of-the-art detection methods, particularly those leveraging deep learning paradigms, encounter a distinct set of challenges when applied to the specific constraints of industrial manufacturing environments. The first challenge is the significant computational burden associated with high-precision models. Accurately detecting minute defects typically necessitates high-resolution imagery; however, processing such data with deep convolutional neural networks (CNNs) or complex Transformer models consumes substantial computational resources and memory. This computational demand often translates to prohibitive hardware costs for widespread deployment on factory floors. The second challenge is scale variation, which presents a persistent hurdle. PCB defects manifest in widely varying sizes, from tiny pinholes barely visible to the naked eye to larger structural errors like open circuits. A detection model must possess the multi-scale capability to identify these diverse anomalies with uniform accuracy. The third challenge is the imperative for real-time performance. In high-velocity production lines, inspection algorithms must operate at speeds commensurate with manufacturing throughput to avoid creating bottlenecks. Unfortunately, many existing high-accuracy models are computationally too intensive for real-time inference, forcing manufacturers to make an undesirable trade-off between inspection speed and detection precision. These limitations highlight the need for an optimized architecture that balances resource efficiency with robust detection capabilities.

Methods

To address these pressing challenges effectively, this study proposes LightRT-DETR, a lightweight real-time detection transformer specifically optimized for PCB defect detection. Building upon the architectural principles of the real-time detection Transformer (RT-DETR), our approach introduces a series of strategic modifications designed to mitigate computational costs significantly while maintaining, and in some aspects enhancing, detection accuracy. The design of LightRT-DETR is characterized by three primary technical innovations: the implementation of the EfficientFastGLU-18 backbone, the integration of a cascaded group attention (CGA) mechanism, and the development of an efficient selective feature pyramid network (ES-FPN). These components work in concert to refine the model's ability to extract relevant features from complex backgrounds without incurring the heavy computational penalties associated with standard transformer architectures. The overarching goal is to create a model that is both agile enough for edge deployment and accurate enough to meet rigorous quality standards. To resolve the inefficiency of traditional backbones like ResNet-18 when deployed on edge computing devices, we propose the integration of the EfficientFastGLU-18 backbone. This novel structure synthesizes partial convolution (PConv) with convolutional gated linear units (GLU). PConv operates by applying convolution to only a subset of input channels while leaving the remainder untouched for the next layer. This strategy effectively reduces memory access and floating-point operations (FLOPs) without compromising the flow of information. By coupling this efficient convolution strategy with GLU, which provides robust non-linear transformation capabilities, the backbone maintains strong feature extraction abilities. This is particularly crucial for identifying defects against the complex, texture-rich backgrounds of PCBs, where differentiating between a valid trace and a defect requires discerning subtle feature variations. Furthermore, to address the computational expense of standard multi-head attention mechanisms, we introduce a cascaded group attention (CGA) mechanism. Instead of processing all feature channels simultaneously, CGA splits feature maps into distinct groups and processes them through a cascaded structure. This approach allows the network to refine features progressively, as the output of one group's attention computation can inform the next, creating a hierarchical refinement process. This mechanism appears particularly beneficial for defects possessing intricate or fine-grained details, such as "mouse bites" or "spurs", which are frequently overlooked by coarser, global attention mechanisms. Finally, we tackle multi-scale defect detection through the efficient selective feature pyramid network (ES-FPN). This redesigned module employs selective feature integration and incorporates RetBlockC3 modules enhanced with manhattan self-attention (MaSA). Unlike standard self-attention, MaSA is specifically designed to capture long-range dependencies in horizontal and vertical directions. Given that PCB layouts consist largely of orthogonal lines, this domain-specific adaptation enables more effective recognition of defects across widely varying scales by aligning the attention mechanism with the geometric priors of the PCB domain.

Results and Discussions

The efficacy of LightRT-DETR was validated through extensive experiments on a public PCB defect dataset, revealing a significant improvement in both efficiency and accuracy compared to current state-of-the-art models, including the widely utilized YOLO series. In terms of quantitative performance, LightRT-DETR achieved a mean average precision (mAP50) of 98.3% and a stricter metric, mAP50-95, of 52.9%. When benchmarked against the advanced YOLOv9e model, our approach demonstrated an improvement of 3.0% in mAP50 and 1.8% in mAP50-95. These metrics suggest that the model is not only capable of detecting defects but also does so with high localization accuracy, minimizing false positives and false negatives—a critical requirement for industrial quality control. The experimental setup ensured fair comparisons by utilizing consistent hyperparameters and training environments, further reinforcing the validity of the observed performance gains.

Conclusions

Perhaps the most distinct advantage of LightRT-DETR lies in its computational efficiency. The model requires approximately 37.2 GFLOPs, representing an 80.3% reduction in computational cost compared to the baseline RT-DETR model. Similarly, the parameter count is reduced to 12.8 million, a 77.7% decrease. This substantial reduction has practical implications: it facilitates deployment on cost-effective hardware with limited computing power or allows the system to operate at significantly higher frame rates on standard hardware.

, authors=Zhiyuan Xi1, 3, Ting Mei1, 3, Shanling Lin2, 3, Zhixian Lin1, 2, 3, Jianpu Lin2, 3, Shanhong Lü2, 3, Tailiang Guo1, 3, authorsList=Zhiyuan Xi, Ting Mei, Shanling Lin, Zhixian Lin, Jianpu Lin, Shanhong Lü, Tailiang Guo, authorCompany=null, correspAuthors=Shanling Lin, authorNote=null, correspAuthorsNote=
, copyrightStatement=Copyright © 2026 Opto-Electronic Engineering. All rights reserved., copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, fund=null), CN=ArticleExt(id=1319955564260610835, articleId=1319955554672431797, tenantId=1146029695717560320, journalId=1278651732997652489, language=CN, title=基于LightRT-DETR算法的印刷电路板缺陷检测, columnId=1319955120360636489, journalTitle=光电工程, columnName=研究论文, runingTitle=null, highlight=null, articleAbstract=

针对工业生产中印刷电路板(printed circuit board, PCB)缺陷检测精度和实时性不高的问题,本文提出了一种基于LightRT-DETR的PCB缺陷检测算法。首先,设计了一种高效的主干网络EfficientFastGLU-18,该网络通过引入部分卷积(partial convolution, PConv) 和卷积门控线性单元(convolutional gated linear unit, convolutional GLU) 机制来降低ResNet-18架构的计算复杂度和参数量,同时保持较好的特征提取能力。其次,采用级联分组注意力机制替代了RT-DETR中的多头注意力机制,通过特征分割与级联,增强输入特征的多样性,并降低不必要的计算开销。最后,提出了一种高效选择性特征金字塔网络(efficient selective feature pyramid network, ES-FPN),其主要由特征选择模块和串联RetBlockC3的特征融合模块构成,增强对细微缺陷的检测能力的同时减少计算冗余,从而实现更高效的实时缺陷检测。实验结果表明,本文改进的模型的mAP50达到了98.3%,mAP50-95值较基准模型RT-DETR提高了11.1%,GFLOPs下降了34.7%,Params降低了35.3%,FPS提升了约5.1%。这些优化使得LightRT-DETR模型在移动设备或嵌入式系统中更加适用。

, authors=希治远1, 3, 梅婷1, 3, 林珊玲2, 3, *, *, 林志贤1, 2, 3, 林坚普2, 3, 吕珊红2, 3, 郭太良1, 3, authorsList=希治远, 梅婷, 林珊玲, 林志贤, 林坚普, 吕珊红, 郭太良, authorCompany=null, correspAuthors=林珊玲, authorNote=

作者贡献

希治远:研究设计、论文初稿撰写、算法实现与验证;梅婷:参与实验设计与结果分析;林珊玲:负责研究方向把控、论文结构审阅与最终定稿;林志贤,林坚普,吕珊红:协助完成基准对比实验;郭太良:提供实验理论指导。

希治远(2000-),男,硕士研究生,主要研究方向为工业缺陷检测。E-mail:

林珊玲(1991-),女,福建泉州人,博士,讲师,硕士研究生导师,福州大学先进制造学院电子信息系教师,主要从事主要研究新型显示技术、平板显示器驱动系统图像处理技术等。E-mail:

, correspAuthorsNote=
林珊玲,
, copyrightStatement=版权所有©《光电工程》编辑部 2026, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=3ZPWklhpjTLw3Ie8EuIafQ==, magXml=NX5FXwp0lE1p+Wsyoh+GDw==, pdfUrl=null, pdf=Ksc8nCIMS0Rmgu3fJYVz7Q==, pdfFileSize=10567851, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=h35hxoiWB7RZf9bl/T5v7A==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=IK3ULnr5qO7RZRXUhqPB6g==, mapNumber=null, fund=null)}, authors=[Author(id=1319955564759733022, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=xzy_20001222@163.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955564839424801, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955564759733022, language=EN, stringName=Zhiyuan Xi, firstName=Zhiyuan, middleName=null, lastName=Xi, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955564906533666, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955564759733022, language=CN, stringName=希治远, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1福州大学物理与信息工程学院,福建 福州 350116
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio={"img":"EMz6wn9MvDFOZjEb1n37kg==","content":"

希治远(2000-),男,硕士研究生,主要研究方向为工业缺陷检测。E-mail:

"}, bioImg=EMz6wn9MvDFOZjEb1n37kg==, bioContent=

希治远(2000-),男,硕士研究生,主要研究方向为工业缺陷检测。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564478714644, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=1, ext=[AuthorCompanyExt(id=1319955564487103253, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564491297558, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学物理与信息工程学院,福建 福州 350116)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955564990419748, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955565120443175, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955564990419748, language=EN, stringName=Ting Mei, firstName=Ting, middleName=null, lastName=Mei, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955565187552040, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955564990419748, language=CN, stringName=梅婷, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1福州大学物理与信息工程学院,福建 福州 350116
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564478714644, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=1, ext=[AuthorCompanyExt(id=1319955564487103253, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564491297558, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学物理与信息工程学院,福建 福州 350116)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955565246272298, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=sllin@fzu.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1319955565334352685, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565246272298, language=EN, stringName=Shanling Lin, firstName=Shanling, middleName=null, lastName=Lin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955565393072942, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565246272298, language=CN, stringName=林珊玲, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, *, *, address=2福州大学先进制造学院,福建 泉州 362000
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio={"img":"vkVx9jvf4c0HmcQkiDwsyQ==","content":"

林珊玲(1991-),女,福建泉州人,博士,讲师,硕士研究生导师,福州大学先进制造学院电子信息系教师,主要从事主要研究新型显示技术、平板显示器驱动系统图像处理技术等。E-mail:

"}, bioImg=vkVx9jvf4c0HmcQkiDwsyQ==, bioContent=

林珊玲(1991-),女,福建泉州人,博士,讲师,硕士研究生导师,福州大学先进制造学院电子信息系教师,主要从事主要研究新型显示技术、平板显示器驱动系统图像处理技术等。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564566795031, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=2, ext=[AuthorCompanyExt(id=1319955564575183640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China), AuthorCompanyExt(id=1319955564583572249, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学先进制造学院,福建 泉州 362000)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955565455987504, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955565539873588, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565455987504, language=EN, stringName=Zhixian Lin, firstName=Zhixian, middleName=null, lastName=Lin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, 3, address=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China
2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955565598593845, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565455987504, language=CN, stringName=林志贤, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, 3, address=1福州大学物理与信息工程学院,福建 福州 350116
2福州大学先进制造学院,福建 泉州 362000
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564478714644, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=1, ext=[AuthorCompanyExt(id=1319955564487103253, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564491297558, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学物理与信息工程学院,福建 福州 350116)]), AuthorCompany(id=1319955564566795031, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=2, ext=[AuthorCompanyExt(id=1319955564575183640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China), AuthorCompanyExt(id=1319955564583572249, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学先进制造学院,福建 泉州 362000)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955565661508407, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955565737005882, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565661508407, language=EN, stringName=Jianpu Lin, firstName=Jianpu, middleName=null, lastName=Lin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955565804114747, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565661508407, language=CN, stringName=林坚普, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2福州大学先进制造学院,福建 泉州 362000
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564566795031, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=2, ext=[AuthorCompanyExt(id=1319955564575183640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China), AuthorCompanyExt(id=1319955564583572249, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学先进制造学院,福建 泉州 362000)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955565904778045, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=5, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955566001247040, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565904778045, language=EN, stringName=Shanhong Lü, firstName=Shanhong, middleName=null, lastName=Lü, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955566085133121, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565904778045, language=CN, stringName=吕珊红, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2福州大学先进制造学院,福建 泉州 362000
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564566795031, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=2, ext=[AuthorCompanyExt(id=1319955564575183640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China), AuthorCompanyExt(id=1319955564583572249, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学先进制造学院,福建 泉州 362000)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955566152241987, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=6, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955566265488198, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955566152241987, language=EN, stringName=Tailiang Guo, firstName=Tailiang, middleName=null, lastName=Guo, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955566332597063, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955566152241987, language=CN, stringName=郭太良, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1福州大学物理与信息工程学院,福建 福州 350116
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564478714644, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=1, ext=[AuthorCompanyExt(id=1319955564487103253, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564491297558, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学物理与信息工程学院,福建 福州 350116)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])])], keywords=[Keyword(id=1319955566458426184, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, orderNo=1, keyword=defect detection), Keyword(id=1319955566517146441, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, orderNo=2, keyword=printed circuit board), Keyword(id=1319955566601032522, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, orderNo=3, keyword=backbone network), Keyword(id=1319955566680724299, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, orderNo=4, keyword=feature pyramid network), Keyword(id=1319955566756221772, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, orderNo=5, keyword=computational redundancy), Keyword(id=1319955566831719245, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, orderNo=1, keyword=缺陷检测), Keyword(id=1319955566911411022, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, orderNo=2, keyword=印刷电路板), Keyword(id=1319955566999491407, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, orderNo=3, keyword=主干网络), Keyword(id=1319955567116931920, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, orderNo=4, keyword=特征金字塔网络), Keyword(id=1319955567217595217, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, orderNo=5, keyword=计算冗余)], refs=[Reference(id=1319955572812796803, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=1, rfOrder=0, authorNames=null, journalName=null, refType=null, unstructuredReference=Lan Z, Hong Y, Li Y. An improved YOLOv3 method for PCB surface defect detection[C]//2021 IEEE International Conference on Power Electronics, Computer Applications (ICPECA), 2021: 1009–1015. https://doi.org/10.1109/ICPECA51329.2021.9362675., articleTitle=null, refAbstract=null), Reference(id=1319955572917654405, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=2, rfOrder=1, authorNames=null, journalName=null, refType=null, unstructuredReference=Redmon J, Divvala S, Girshick R, et al. You only look once: unified, real-time object detection[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, 2016: 779–788. https://doi.org/10.1109/CVPR.2016.91., articleTitle=null, refAbstract=null), Reference(id=1319955573001540486, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=3, rfOrder=2, authorNames=null, journalName=null, refType=null, unstructuredReference=Lin T Y, Goyal P, Girshick R, et al. Focal loss for dense object detection[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision, 2017: 2999–3007. https://doi.org/10.1109/ICCV.2017.324., articleTitle=null, refAbstract=null), Reference(id=1319955573098009480, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=4, rfOrder=3, authorNames=null, journalName=null, refType=null, unstructuredReference=Liu W, Anguelov D, Erhan D, et al. SSD: single shot MultiBox detector[C]//14th European Conference on Computer Vision, 2016: 21–37. https://doi.org/10.1007/978-3-319-46448-0_2., articleTitle=null, refAbstract=null), Reference(id=1319955573219644298, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=5, rfOrder=4, authorNames=null, journalName=null, refType=null, unstructuredReference=Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014: 580–587. https://doi.org/10.1109/CVPR.2014.81., articleTitle=null, refAbstract=null), Reference(id=1319955573358056332, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=6, rfOrder=5, authorNames=null, journalName=null, refType=null, unstructuredReference=Girshick R. Fast R-CNN[C]//Proceedings of the 2015 IEEE International Conference on Computer Vision, 2015: 1440–1448. https://doi.org/10.1109/ICCV.2015.169., articleTitle=null, refAbstract=null), Reference(id=1319955573475496846, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=7, rfOrder=6, authorNames=null, journalName=null, refType=null, unstructuredReference=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 29th International Conference on Neural Information Processing Systems, 2015: 91–99., articleTitle=null, refAbstract=null), Reference(id=1319955573563577232, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=8, rfOrder=7, authorNames=null, journalName=null, refType=null, unstructuredReference=He K M, Gkioxari G, Dollár P, et al. Mask R-CNN[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision, 2017: 2980–2988. https://doi.org/10.1109/ICCV.2017.322., articleTitle=null, refAbstract=null), Reference(id=1319955573672629138, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=9, rfOrder=8, authorNames=null, journalName=null, refType=null, unstructuredReference=Lian J, Wang L T, Liu T Y, et al. Automatic visual inspection for printed circuit board via novel Mask R-CNN in smart city applications[J]. Sustain Energy Technol Assess, 2021, 44: 101032., articleTitle=null, refAbstract=null), Reference(id=1319955573773292436, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=10, rfOrder=9, authorNames=null, journalName=null, refType=null, unstructuredReference=Li Y T, Kuo P, Guo J I. Automatic industry PCB board DIP process defect detection with deep ensemble method[C]//2020 IEEE 29th International Symposium on Industrial Electronics (ISIE), 2020: 453–459. https://doi.org/10.1109/ISIE45063.2020.9152533., articleTitle=null, refAbstract=null), Reference(id=1319955573903315862, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=11, rfOrder=10, authorNames=null, journalName=null, refType=null, unstructuredReference=Xin H J, Chen Z B, Wang B Y. PCB electronic component defect detection method based on improved YOLOv4 algorithm[J]. J Phys Conf Ser, 2021, 1827(1): 012167., articleTitle=null, refAbstract=null), Reference(id=1319955574020756376, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=12, rfOrder=11, authorNames=null, journalName=null, refType=null, unstructuredReference=Xie H F, Li Y, Li X C, et al. A method for surface defect detection of printed circuit board based on improved YOLOv4[C]//2021 IEEE 2nd International Conference on Big Data, Artificial Intelligence and Internet of Things Engineering (ICBAIE), 2021: 851–857. https://doi.org/10.1109/ICBAIE52039.2021.9390006., articleTitle=null, refAbstract=null), Reference(id=1319955574117225370, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=13, rfOrder=12, authorNames=null, journalName=null, refType=null, unstructuredReference=Wu L G, Zhang L, Zhou Q. Printed circuit board quality detection method integrating lightweight network and dual attention mechanism[J]. IEEE Access, 2022, 10: 87617−87629., articleTitle=null, refAbstract=null), Reference(id=1319955574255637402, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=14, rfOrder=13, authorNames=null, journalName=null, refType=null, unstructuredReference=Wang X, Gao J S, Hou B J, et al. A lightweight modified YOLOX network using coordinate attention mechanism for PCB surface defect detection[J]. IEEE Sens J, 2022, 22(21): 20910−20920., articleTitle=null, refAbstract=null), Reference(id=1319955574402438045, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=15, rfOrder=14, authorNames=null, journalName=null, refType=null, unstructuredReference=陶志勇, 何燕, 林森, 等. 局部和全局特征融合的太阳能电池片表面缺陷检测[J]. 光电工程, 2024, 51(1): 230292., articleTitle=null, refAbstract=null), Reference(id=1319955574524072863, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=15, rfOrder=15, authorNames=null, journalName=null, refType=null, unstructuredReference=Tao Z Y, He Y, Lin S, et al. Surface defect detection of solar cells using local and global feature fusion[J]. Opto-Electron Eng, 2024, 51(1): 230292., articleTitle=null, refAbstract=null), Reference(id=1319955574633124769, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=16, rfOrder=16, authorNames=null, journalName=null, refType=null, unstructuredReference=Carion N, Massa F, Synnaeve G, et al. End-to-end object detection with transformers[C]//16th European Conference on Computer Vision, 2020: 213–229. https://doi.org/10.1007/978-3-030-58452-8_13., articleTitle=null, refAbstract=null), Reference(id=1319955574788314019, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=17, rfOrder=17, authorNames=null, journalName=null, refType=null, unstructuredReference=Zhao Y, Lv W Y, Xu S L, et al. DETRs beat YOLOs on real-time object detection[C]//Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 16965–16974. https://doi.org/10.1109/CVPR52733.2024.01605., articleTitle=null, refAbstract=null), Reference(id=1319955574905754533, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=18, rfOrder=18, authorNames=null, journalName=null, refType=null, unstructuredReference=He K M, Zhang X Y, Ren A Q, et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770–778. https://doi.org/10.1109/CVPR.2016.90., articleTitle=null, refAbstract=null), Reference(id=1319955575006417831, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=19, rfOrder=19, authorNames=null, journalName=null, refType=null, unstructuredReference=Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems, 2017: 6000–6010., articleTitle=null, refAbstract=null), Reference(id=1319955575102886825, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=20, rfOrder=20, authorNames=null, journalName=null, refType=null, unstructuredReference=Liu X Y, Peng H W, Zheng N X, et al. EfficientViT: memory efficient vision transformer with cascaded group attention[C]//Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023: 14420–14430. https://doi.org/10.1109/CVPR52729.2023.01386., articleTitle=null, refAbstract=null), Reference(id=1319955575211938731, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=21, rfOrder=21, authorNames=null, journalName=null, refType=null, unstructuredReference=Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 7132–7141. https://doi.org/10.1109/CVPR.2018.00745., articleTitle=null, refAbstract=null), Reference(id=1319955575270658989, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=22, rfOrder=22, authorNames=null, journalName=null, refType=null, unstructuredReference=Fan Q H, Huang H B, Chen M R, et al. RMT: retentive networks meet vision transformers[C]//Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 5641–5651. https://doi.org/10.1109/CVPR52733.2024.00539., articleTitle=null, refAbstract=null), Reference(id=1319955575354545071, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=23, rfOrder=23, authorNames=null, journalName=null, refType=null, unstructuredReference=吴斌斌, 张礼华, 刘军伟, 等. 基于改进的EP-RTDETR小目标PCB表面缺陷检测[J]. 制造技术与机床, 2025, (3): 139−148., articleTitle=null, refAbstract=null), Reference(id=1319955575430042545, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=23, rfOrder=24, authorNames=null, journalName=null, refType=null, unstructuredReference=Wu B B, Zhang L H, Liu J W, et al. Improved EP-RTDETR based surface defect detection on PCB[J]. Manuf Technol Mach Tool, 2025, (3): 139−148., articleTitle=null, refAbstract=null), Reference(id=1319955575518122930, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=24, rfOrder=25, authorNames=null, journalName=null, refType=null, unstructuredReference=陈仁祥, 詹赞, 胡小林, 等. 基于多注意力Faster RCNN的噪声干扰下印刷电路板缺陷检测[J]. 仪器仪表学报, 2021, 42(12): 167−174., articleTitle=null, refAbstract=null), Reference(id=1319955575597814708, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=24, rfOrder=26, authorNames=null, journalName=null, refType=null, unstructuredReference=Chen R X, Zhan Z, Hu X L, et al. Printed circuit board defect detection based on the multi-attentive faster RCNN under noise interference[J]. Chin J Sci Instrum, 2021, 42(12): 167−174., articleTitle=null, refAbstract=null), Reference(id=1319955575677506486, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=25, rfOrder=27, authorNames=null, journalName=null, refType=null, unstructuredReference=Luo S, Xu Y P, Zhang C L, et al. LIDD-YOLO: a lightweight industrial defect detection network[J]. Meas Sci Technol, 2024, 36(1): 0161b5., articleTitle=null, refAbstract=null), Reference(id=1319955575782364086, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=26, rfOrder=28, authorNames=null, journalName=null, refType=null, unstructuredReference=陈建豪, 徐洁, 汪志锋. 基于深度学习的PCB缺陷检测[J]. 现代电子技术, 2025, 48(8): 7−12., articleTitle=null, refAbstract=null), Reference(id=1319955575862055863, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=26, rfOrder=29, authorNames=null, journalName=null, refType=null, unstructuredReference=Chen J H, Xu J, Wang Z F. PCB defect detection based on deep learning[J]. Mod Electron Tech, 2025, 48(8): 7−12., articleTitle=null, refAbstract=null), Reference(id=1319955575924970425, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=27, rfOrder=30, authorNames=null, journalName=null, refType=null, unstructuredReference=Wang H D, Xie J, Xu X Y, et al. Few-shot PCB surface defect detection based on feature enhancement and multi-scale fusion[J]. IEEE Access, 2022, 10: 129911−129924., articleTitle=null, refAbstract=null), Reference(id=1319955576000467899, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=28, rfOrder=31, authorNames=null, journalName=null, refType=null, unstructuredReference=Zhang L X, Chen J S, Chen J G, et al. LDD-Net: lightweight printed circuit board defect detection network fusing multi-scale features[J]. Eng Appl Artif Intell, 2024, 129: 107628., articleTitle=null, refAbstract=null), Reference(id=1319955576088548285, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=29, rfOrder=32, authorNames=null, journalName=null, refType=null, unstructuredReference=Li C J, Qu Z, Wang S Y, et al. A method of defect detection for focal hard samples PCB based on extended FPN model[J]. IEEE Trans Compon Packaging Manuf Technol, 2022, 12(2): 217−227., articleTitle=null, refAbstract=null), Reference(id=1319955576155657150, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=30, rfOrder=33, authorNames=null, journalName=null, refType=null, unstructuredReference=Shao R, Zhou M L, Li M, et al. TD-Net: tiny defect detection network for industrial products[J]. Complex Intell Syst, 2024, 10(3): 3943−3954., articleTitle=null, refAbstract=null)], funds=null, companyList=[AuthorCompany(id=1319955564478714644, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=1, ext=[AuthorCompanyExt(id=1319955564487103253, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564491297558, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学物理与信息工程学院,福建 福州 350116)]), AuthorCompany(id=1319955564566795031, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=2, ext=[AuthorCompanyExt(id=1319955564575183640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China), AuthorCompanyExt(id=1319955564583572249, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学先进制造学院,福建 泉州 362000)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])], figs=[ArticleFig(id=1319955567372784466, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.1, caption=Network architecture of the RT-DETR-R18 model, figureFileSmall=8kgB39lFmtksFciMh/FhzA==, figureFileBig=h35hxoiWB7RZf9bl/T5v7A==, tableContent=null), ArticleFig(id=1319955567452476243, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图1, caption=RT-DETR-R18 模型网络结构, figureFileSmall=8kgB39lFmtksFciMh/FhzA==, figureFileBig=h35hxoiWB7RZf9bl/T5v7A==, tableContent=null), ArticleFig(id=1319955567662191444, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.2, caption=Network architecture of the LightRT-DETR model, figureFileSmall=Xs7ZnZZ28/Lq2fVhs6ZwNw==, figureFileBig=yqcvC+lMJhxCWuWaC3hZhw==, tableContent=null), ArticleFig(id=1319955567754466133, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图2, caption=LightRT-DETR模型网络结构, figureFileSmall=Xs7ZnZZ28/Lq2fVhs6ZwNw==, figureFileBig=yqcvC+lMJhxCWuWaC3hZhw==, tableContent=null), ArticleFig(id=1319955567859323734, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.3, caption=EfficientFastGLU architecture, figureFileSmall=PE01KO5LOR1XKA5b3OrljQ==, figureFileBig=QRgZ+F6PpEY0IHwcXT1cIA==, tableContent=null), ArticleFig(id=1319955567939015511, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图3, caption=EfficientFastGLU架构, figureFileSmall=PE01KO5LOR1XKA5b3OrljQ==, figureFileBig=QRgZ+F6PpEY0IHwcXT1cIA==, tableContent=null), ArticleFig(id=1319955568035484504, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.4, caption=Network structure diagram of Pconv, figureFileSmall=hbMCixbmtA8jz9Ck+Rff/w==, figureFileBig=n8QPryYPWq6ZTTYk75rWkw==, tableContent=null), ArticleFig(id=1319955568123564889, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图4, caption=Pconv网络结构图, figureFileSmall=hbMCixbmtA8jz9Ck+Rff/w==, figureFileBig=n8QPryYPWq6ZTTYk75rWkw==, tableContent=null), ArticleFig(id=1319955568203256666, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.5, caption=Structural comparison between Convolutional GLU module and ConvFFN module. (a) Convolutional GLU module; (b) ConvFFN module, figureFileSmall=eSQerNyAQdYv0cuogkiy1g==, figureFileBig=YSFNh231mnCuGOWkCPPX5w==, tableContent=null), ArticleFig(id=1319955568270365531, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图5, caption=Convolutional GLU模块与ConvFFN模块结构对比图。(a) Convolutional GLU模块;(b) ConvFFN模块, figureFileSmall=eSQerNyAQdYv0cuogkiy1g==, figureFileBig=YSFNh231mnCuGOWkCPPX5w==, tableContent=null), ArticleFig(id=1319955568337474396, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.6, caption=CGA attention mechanism, figureFileSmall=wkMWY28IeV+kDN6OdQcB5Q==, figureFileBig=D8r0gbiE9EmQ+ImTAt7kKg==, tableContent=null), ArticleFig(id=1319955568421360477, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图6, caption=CGA注意力机制, figureFileSmall=wkMWY28IeV+kDN6OdQcB5Q==, figureFileBig=D8r0gbiE9EmQ+ImTAt7kKg==, tableContent=null), ArticleFig(id=1319955568501052254, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.7, caption=ES-FPN architecture, figureFileSmall=mPzNRZ3f1+6aiFNR6Mp2BQ==, figureFileBig=e4TOSLCJJkmSJVV/JMM6MA==, tableContent=null), ArticleFig(id=1319955568584938335, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图7, caption=ES-FPN架构, figureFileSmall=mPzNRZ3f1+6aiFNR6Mp2BQ==, figureFileBig=e4TOSLCJJkmSJVV/JMM6MA==, tableContent=null), ArticleFig(id=1319955568660435808, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.8, caption=CA Module architecture, figureFileSmall=o+QD/Sl8m0nVZ0HidURgiQ==, figureFileBig=xuI+KC1k+G8rKYO2Ruq5yw==, tableContent=null), ArticleFig(id=1319955568756904801, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图8, caption=CA模块架构, figureFileSmall=o+QD/Sl8m0nVZ0HidURgiQ==, figureFileBig=xuI+KC1k+G8rKYO2Ruq5yw==, tableContent=null), ArticleFig(id=1319955568878539618, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.9, caption=RetBlockC3 module architecture, figureFileSmall=k6o2YdpCK9zry4K78PdbdQ==, figureFileBig=2VNxwm9+YLyg4QOZD54GtA==, tableContent=null), ArticleFig(id=1319955569000174435, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图9, caption=RetBlockC3模块架构, figureFileSmall=k6o2YdpCK9zry4K78PdbdQ==, figureFileBig=2VNxwm9+YLyg4QOZD54GtA==, tableContent=null), ArticleFig(id=1319955569071477604, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.10, caption=Comparison of loss values and mAP images before and after incorporating PConv. (a) Comparison image of loss value changes; (b) Comparison image of mAP50 changes, figureFileSmall=T0MdSv9ZPZTpo87ewUaOrg==, figureFileBig=oW/Xk548h5jJf7KB/feSjg==, tableContent=null), ArticleFig(id=1319955569205695334, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图10, caption=引入PConv前后的损失值图像与mAP图像对比。(a)损失值变化的对比图;(b) mAP50变化的对比图, figureFileSmall=T0MdSv9ZPZTpo87ewUaOrg==, figureFileBig=oW/Xk548h5jJf7KB/feSjg==, tableContent=null), ArticleFig(id=1319955569285387111, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.11, caption=Performance and feature analysis before and after introducing PConv. (a) Comparison bar chart of mAP50 changes; (b) Feature space similarity between RT-DETR-R18 and EfficientFastGLU-18; (c) Feature spectral energy comparison between RT-DETR-R18 and EfficientFastGLU-18, figureFileSmall=DhTXvHemAcGeb2LlbdoGiQ==, figureFileBig=amTTm2M2mrpcUTv+R/Sp8A==, tableContent=null), ArticleFig(id=1319955569390244712, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图11, caption=引入PConv前后的性能与特征分析。(a) mAP50变化的对比柱状图;(b) RT-DETR-R18与EfficientFastGLU-18的特征空间相似性;(c) RT-DETR-R18与EfficientFastGLU-18的特征谱能量对比, figureFileSmall=DhTXvHemAcGeb2LlbdoGiQ==, figureFileBig=amTTm2M2mrpcUTv+R/Sp8A==, tableContent=null), ArticleFig(id=1319955569478325097, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.12, caption=Comparison of loss values. (a) Curves of loss value versus epoch; (b) Curves of mAP50 versus epoch, figureFileSmall=Nq5UHoqdSs3l5ennd+zXNw==, figureFileBig=cdmLYqoaTUGC0lIYNbIvmA==, tableContent=null), ArticleFig(id=1319955569562211178, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图12, caption=损失值图像对比。(a)损失值随epoch变化曲线;(b) mAP50随epoch变化曲线, figureFileSmall=Nq5UHoqdSs3l5ennd+zXNw==, figureFileBig=cdmLYqoaTUGC0lIYNbIvmA==, tableContent=null), ArticleFig(id=1319955569637708651, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.13, caption=Comparison of detection results before and after improvement. (a-f) Represent the original annotations and model predictions results for missing hole, mouse bite, open circuit, short circuit, spur, and spurious copper, respectively, figureFileSmall=5JAFivltbmN0lbAAP/VcfQ==, figureFileBig=TEL9GNoFllB+MytBxShNAw==, tableContent=null), ArticleFig(id=1319955569713206124, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图13, caption=改进前后检测结果对比。(a~f)分别代表漏孔、鼠咬、开路、短路、杂散和杂铜的原始标注及模型推理效果, figureFileSmall=5JAFivltbmN0lbAAP/VcfQ==, figureFileBig=TEL9GNoFllB+MytBxShNAw==, tableContent=null), ArticleFig(id=1319955569780314989, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Fig.14, caption=Comparison of heatmaps before and after improvement. (a) Original annotation image; (b) RT-DETR heatmap; (c) LightRT-DETR heatmap, figureFileSmall=80hRFpf80EOyVBzuxehaYQ==, figureFileBig=PvMVIQZeJHezgIFO3G+RDQ==, tableContent=null), ArticleFig(id=1319955569860006766, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=图14, caption=改进前后热力图对比。(a)原始标注图;(b) RT-DETR热力图;(c) LightRT-DETR热力图, figureFileSmall=80hRFpf80EOyVBzuxehaYQ==, figureFileBig=PvMVIQZeJHezgIFO3G+RDQ==, tableContent=null), ArticleFig(id=1319955569922921327, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Tab.1, caption=

Dataset information

, figureFileSmall=null, figureFileBig=null, tableContent=
Defect typeNumber of imagesNumber of defects
Missing hole115449
Mouse bite115411
Open circuit116453
Short circuit116433
Spurious copper116473
Spur115443
), ArticleFig(id=1319955569985835888, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=表1, caption=

数据集信息

, figureFileSmall=null, figureFileBig=null, tableContent=
Defect typeNumber of imagesNumber of defects
Missing hole115449
Mouse bite115411
Open circuit116453
Short circuit116433
Spurious copper116473
Spur115443
), ArticleFig(id=1319955570044556145, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Tab.2, caption=

Comparison of improvement strategies for backbone networks

, figureFileSmall=null, figureFileBig=null, tableContent=
Base modelBackbone network improvementmAP50/%GFLOPsFPSParams/M
RT-DETRResnet18 (baseline)96.257.056.419.8
RT-DETREfficientFormer_v295.729.544.711.8
RT-DETREfficientViT90.027.232.110.7
RT-DETRRepViT88.436.445.54.4
RT-DETRFasternet94.428.564.910.8
RT-DETRMobileNet_v491.539.564.011.3
RT-DETREfficientFastGLU-1896.946.257.215.4
), ArticleFig(id=1319955570128442226, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=表2, caption=

主干网络的改进策略对比

, figureFileSmall=null, figureFileBig=null, tableContent=
Base modelBackbone network improvementmAP50/%GFLOPsFPSParams/M
RT-DETRResnet18 (baseline)96.257.056.419.8
RT-DETREfficientFormer_v295.729.544.711.8
RT-DETREfficientViT90.027.232.110.7
RT-DETRRepViT88.436.445.54.4
RT-DETRFasternet94.428.564.910.8
RT-DETRMobileNet_v491.539.564.011.3
RT-DETREfficientFastGLU-1896.946.257.215.4
), ArticleFig(id=1319955570191356787, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Tab.3, caption=

The results comparison of EfficientFastGLU-18 with different channel reduction factors r

, figureFileSmall=null, figureFileBig=null, tableContent=
Reduction ratio rmAP50/%GFLOPsFPS
0.12595.745.558.8
0.25096.946.257.2
0.50096.949.053.4
), ArticleFig(id=1319955570359128948, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=表3, caption=

EfficientFastGLU-18采用不同的通道缩减因子r结果对比

, figureFileSmall=null, figureFileBig=null, tableContent=
Reduction ratio rmAP50/%GFLOPsFPS
0.12595.745.558.8
0.25096.946.257.2
0.50096.949.053.4
), ArticleFig(id=1319955570451403637, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Tab.4, caption=

Comparison of attention mechanism improvements

, figureFileSmall=null, figureFileBig=null, tableContent=
Base modelAttentionmAP50/%GFLOPsFPSParams/M
RT-DETRMHA attention (baseline)96.257.056.419.8
RT-DETRFocused linear attention94.956.756.719.2
RT-DETRPerformer attention95.756.758.719.3
RT-DETRLinformer attention95.655.760.319.3
RT-DETRHiLo attention94.857.153.419.8
RT-DETRDeformable attention transformer95.257.255.019.8
RT-DETRCGA attention97.757.057.819.7
), ArticleFig(id=1319955570833085302, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=表4, caption=

注意力机制的改进策略对比

, figureFileSmall=null, figureFileBig=null, tableContent=
Base modelAttentionmAP50/%GFLOPsFPSParams/M
RT-DETRMHA attention (baseline)96.257.056.419.8
RT-DETRFocused linear attention94.956.756.719.2
RT-DETRPerformer attention95.756.758.719.3
RT-DETRLinformer attention95.655.760.319.3
RT-DETRHiLo attention94.857.153.419.8
RT-DETRDeformable attention transformer95.257.255.019.8
RT-DETRCGA attention97.757.057.819.7
), ArticleFig(id=1319955571206378359, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Tab.5, caption=

Comparison of feature neck improvement strategies

, figureFileSmall=null, figureFileBig=null, tableContent=
Base modelNeck network improvementmAP50/%GFLOPsFPSParams/M
RT-DETRPANet (baseline)96.257.056.419.8
RT-DETRBIFPN95.564.341.420.3
RT-DETRHSFPN96.453.350.518.1
RT-DETRSDFM95.861.944.421.3
RT-DETRPSFM95.271.237.722.4
RT-DETRES-FPN97.447.957.017.4
), ArticleFig(id=1319955571554505592, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=表5, caption=

特征颈部的改进策略对比

, figureFileSmall=null, figureFileBig=null, tableContent=
Base modelNeck network improvementmAP50/%GFLOPsFPSParams/M
RT-DETRPANet (baseline)96.257.056.419.8
RT-DETRBIFPN95.564.341.420.3
RT-DETRHSFPN96.453.350.518.1
RT-DETRSDFM95.861.944.421.3
RT-DETRPSFM95.271.237.722.4
RT-DETRES-FPN97.447.957.017.4
), ArticleFig(id=1319955571676140409, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Tab.6, caption=

Ablation experiment results

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelEfficientFastGLU-18CGA encoderES-FPNmAP50/%mAP50-95/%GFLOPsFPSParams/M
M1×××96.247.657.056.419.8
M2√××96.951.846.257.215.4
M3×√×97.752.457.057.819.7
M4××√97.452.447.957.017.4
M5√×√97.653.037.258.712.9
M6√√×96.952.046.358.015.2
M7×√√97.852.848.058.417.2
M8√√√98.352.937.259.312.8
), ArticleFig(id=1319955571818746747, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=表6, caption=

消融实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelEfficientFastGLU-18CGA encoderES-FPNmAP50/%mAP50-95/%GFLOPsFPSParams/M
M1×××96.247.657.056.419.8
M2√××96.951.846.257.215.4
M3×√×97.752.457.057.819.7
M4××√97.452.447.957.017.4
M5√×√97.653.037.258.712.9
M6√√×96.952.046.358.015.2
M7×√√97.852.848.058.417.2
M8√√√98.352.937.259.312.8
), ArticleFig(id=1319955572049433469, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Tab.7, caption=

Comparison of detection accuracy of each model for different target sizes

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelEfficientFastGLU-18CGA encoderES-FPNmAP50/%
(large target)
mAP50-95/%
(large target)
mAP50/%
(small target)
mAP50-95/%
(small target)
M1×××96.852.396.047.5
M2√××97.653.996.851.8
M3×√×98.454.697.552.3
M4××√98.053.097.252.2
M5√×√98.354.797.552.9
M6√√×97.354.396.952.0
M7×√√98.456.097.852.6
M8√√√98.555.698.252.4
), ArticleFig(id=1319955572145902463, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=表7, caption=

各模型在不同目标尺寸下的精度对比

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelEfficientFastGLU-18CGA encoderES-FPNmAP50/%
(large target)
mAP50-95/%
(large target)
mAP50/%
(small target)
mAP50-95/%
(small target)
M1×××96.852.396.047.5
M2√××97.653.996.851.8
M3×√×98.454.697.552.3
M4××√98.053.097.252.2
M5√×√98.354.797.552.9
M6√√×97.354.396.952.0
M7×√√98.456.097.852.6
M8√√√98.555.698.252.4
), ArticleFig(id=1319955572410143616, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=EN, label=Tab.8, caption=

Comparative experimental results

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelmAP50/%mAP50-95/%GFLOPsParams/M
EP-RTDETR[23]96.746.2—17.7
Improved Faster R-CNN [24]92.9———
LIDD-YOLO[25]93.3—7.12.6
WiseYOLOv7[26]96.157.9136.857
FPFM[27]78.86———
LDD-Net[28]95.90——5.13
ExtendedFPN[29]96.20——137.00
TD-Net[30]96.20——7.06
YOLOv5l94.048.5134.753.1
YOLOv8m91.346.678.725.8
YOLOv8l94.749.3164.843.6
YOLOv9m93.447.876.520.0
YOLOv9e95.351.1189.257.3
LightRT-DETR98.352.937.212.8
), ArticleFig(id=1319955572506612609, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, language=CN, label=表8, caption=

对比实验结果

, figureFileSmall=null, figureFileBig=null, tableContent=
ModelmAP50/%mAP50-95/%GFLOPsParams/M
EP-RTDETR[23]96.746.2—17.7
Improved Faster R-CNN [24]92.9———
LIDD-YOLO[25]93.3—7.12.6
WiseYOLOv7[26]96.157.9136.857
FPFM[27]78.86———
LDD-Net[28]95.90——5.13
ExtendedFPN[29]96.20——137.00
TD-Net[30]96.20——7.06
YOLOv5l94.048.5134.753.1
YOLOv8m91.346.678.725.8
YOLOv8l94.749.3164.843.6
YOLOv9m93.447.876.520.0
YOLOv9e95.351.1189.257.3
LightRT-DETR98.352.937.212.8
)], attaches=null, journal=Journal(id=1278641367198941188, delFlag=0, nameCn=光电工程, nameEn=Opto-Electronic Engineering, nameHistory1=null, nameHistory2=null, issn=1003-501X, eissn=2097-4019, cn=51-1346/O4, coden=null, periodic=0, language=CN, oaType=null, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=4Vimkd+qXLWNxtdpr9mFNw==, journalPrice=null, startedYear=null, abbrevIsoEn=Opto-Electronic Engineering, journalRemark=null, publicationField=null, createdTime=1782781457950, updatedTime=1784021784694, createdBy=18614031015, updatedBy=13041195026, firstLetterCn=G, firstLetterEn=G, subjectCode=Engineering, subjectName=null, subjectCodeEn=Engineering, subjectNameEn=null, picCn=4Vimkd+qXLWNxtdpr9mFNw==, picEn=vAq9s20WLs1ODDfbWq+Gjg==, jcr=null, cjcr=null, exts=[JournalExt(id=1283843675721536154, language=CN, name=光电工程, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1784021784954, updatedTime=1784021784954, createdBy=13041195026, updatedBy=13041195026, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.manuscripts.com.cn/gdgc, submissionEditorUrl=http://www.manuscripts.com.cn/gdgc, submissionReviewUrl=http://www.manuscripts.com.cn/gdgc, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1283843675771867803, language=EN, name=Opto-Electronic Engineering, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1784021784966, updatedTime=1784021784966, createdBy=13041195026, updatedBy=13041195026, submissionGuidelinesUrl=, submissionAuthorUrl=http://www.manuscripts.com.cn/gdgc, submissionEditorUrl=http://www.manuscripts.com.cn/gdgc, submissionReviewUrl=http://www.manuscripts.com.cn/gdgc, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1278651732997652489, websiteList=[Website(id=1278723867418018151, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1278651732997652489, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/oee/CN, language=CN, createTime=1782801127533, createBy=18614031015, updateTime=1782804494442, updateBy=18614031015, name=光电工程-中文, tplId=1146099689490845704, title=光电工程, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1278738091150128034, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=articleTextType, value=kx, createTime=1782804518735, updateTime=1782804518735, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091120767903, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=banner, value=null, createTime=1782804518728, updateTime=1782804518728, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091171099557, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=grayFlag, value=0, createTime=1782804518740, updateTime=1782804518740, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091108184990, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=logo, value=https://castjournals.cast.org.cn/joweb/oee/CN/file/pic?fileId=A1C6uwqtMazluiWkEpR0Mg==, createTime=1782804518725, updateTime=1782804518725, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091179488167, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=minRunFlag, value=0, createTime=1782804518742, updateTime=1782804518742, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091141739425, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/oee/CN/file/pic, createTime=1782804518733, updateTime=1782804518733, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091175293862, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=silenceFlag, value=0, createTime=1782804518741, updateTime=1782804518741, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091129156512, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1782804518730, updateTime=1782804518730, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091154322339, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=themeColor, value=null, createTime=1782804518736, updateTime=1782804518736, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738091162710948, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867418018151, code=themeStyle, value=null, createTime=1782804518738, updateTime=1782804518738, creator=18614031015, updator=18614031015)]), Website(id=1278723867522875769, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1278651732997652489, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/oee/EN, language=EN, createTime=1782801127558, createBy=18614031015, updateTime=1782804490442, updateBy=18614031015, name=光电工程-英文, tplId=1146101810881728533, title=Opto-Electronic Engineering, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1278738063660659607, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=articleTextType, value=kx, createTime=1782804512181, updateTime=1782804512181, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063635493780, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=banner, value=null, createTime=1782804512175, updateTime=1782804512175, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063924900762, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=grayFlag, value=0, createTime=1782804512244, updateTime=1782804512244, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063606133651, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=logo, value=https://castjournals.cast.org.cn/joweb/oee/EN/file/pic?fileId=A1C6uwqtMazluiWkEpR0Mg==, createTime=1782804512168, updateTime=1782804512168, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063937483676, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=minRunFlag, value=0, createTime=1782804512247, updateTime=1782804512247, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063652270998, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/oee/EN/file/pic, createTime=1782804512179, updateTime=1782804512179, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063933289371, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=silenceFlag, value=0, createTime=1782804512246, updateTime=1782804512246, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063639688085, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1782804512176, updateTime=1782804512176, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063664853912, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=themeColor, value=null, createTime=1782804512182, updateTime=1782804512182, creator=18614031015, updator=18614031015), WebsiteProps(id=1278738063916512153, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1278723867522875769, code=themeStyle, value=null, createTime=1782804512242, updateTime=1782804512242, creator=18614031015, updator=18614031015)])], journalTitle=光电工程, weixinUrl=null, journalUrl=https://www.oejournal.org/oee, iacademicId=null, status=1, seqNo=null, journalTitleEn=Opto-Electronic Engineering, journalPhotoCn=4Vimkd+qXLWNxtdpr9mFNw==, journalPhotoEn=vAq9s20WLs1ODDfbWq+Gjg==, journalFirstLetter=G, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false, interPubPlatform=, interPubPlatformUrl=null), detailUrlCn=https://castjournals.cast.org.cn/joweb/oee/CN/10.12086/oee.2026.250117, detailUrlEn=https://castjournals.cast.org.cn/joweb/oee/EN/10.12086/oee.2026.250117, pdfUrlCn=https://castjournals.cast.org.cn/joweb/oee/CN/PDF/10.12086/oee.2026.250117, pdfUrlEn=https://castjournals.cast.org.cn/joweb/oee/EN/PDF/10.12086/oee.2026.250117, aliStartDate=0, aliEndDate=0, collectionFlag=false, citedCount=null, citedUrl=null, previewStatus=0, delFlag=0, hasFullText=1, orderTime=1771948800000, fullTextJson=null, articleText=null, reference=null)
收藏切换
基于LightRT-DETR算法的印刷电路板缺陷检测
收藏切换
PDF下载
希治远 1, 3 , 梅婷 1, 3 , 林珊玲 2, 3, *, * , 林志贤 1, 2, 3 , 林坚普 2, 3 , 吕珊红 2, 3 , 郭太良 1, 3
光电工程 | 研究论文 2026,53(2): 250117-1
收起
收藏切换
光电工程 |研究论文 2026 , 53 (2) : 250117 -1
基于LightRT-DETR算法的印刷电路板缺陷检测
全屏
[Author(id=1319955564759733022, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=xzy_20001222@163.com, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955564839424801, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955564759733022, language=EN, stringName=Zhiyuan Xi, firstName=Zhiyuan, middleName=null, lastName=Xi, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955564906533666, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955564759733022, language=CN, stringName=希治远, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1福州大学物理与信息工程学院,福建 福州 350116
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio={"img":"EMz6wn9MvDFOZjEb1n37kg==","content":"

希治远(2000-),男,硕士研究生,主要研究方向为工业缺陷检测。E-mail:

"}, bioImg=EMz6wn9MvDFOZjEb1n37kg==, bioContent=

希治远(2000-),男,硕士研究生,主要研究方向为工业缺陷检测。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564478714644, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=1, ext=[AuthorCompanyExt(id=1319955564487103253, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564491297558, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学物理与信息工程学院,福建 福州 350116)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955564990419748, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955565120443175, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955564990419748, language=EN, stringName=Ting Mei, firstName=Ting, middleName=null, lastName=Mei, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955565187552040, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955564990419748, language=CN, stringName=梅婷, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1福州大学物理与信息工程学院,福建 福州 350116
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564478714644, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=1, ext=[AuthorCompanyExt(id=1319955564487103253, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564491297558, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学物理与信息工程学院,福建 福州 350116)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955565246272298, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=sllin@fzu.edu.cn, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1319955565334352685, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565246272298, language=EN, stringName=Shanling Lin, firstName=Shanling, middleName=null, lastName=Lin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955565393072942, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565246272298, language=CN, stringName=林珊玲, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, *, *, address=2福州大学先进制造学院,福建 泉州 362000
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio={"img":"vkVx9jvf4c0HmcQkiDwsyQ==","content":"

林珊玲(1991-),女,福建泉州人,博士,讲师,硕士研究生导师,福州大学先进制造学院电子信息系教师,主要从事主要研究新型显示技术、平板显示器驱动系统图像处理技术等。E-mail:

"}, bioImg=vkVx9jvf4c0HmcQkiDwsyQ==, bioContent=

林珊玲(1991-),女,福建泉州人,博士,讲师,硕士研究生导师,福州大学先进制造学院电子信息系教师,主要从事主要研究新型显示技术、平板显示器驱动系统图像处理技术等。E-mail:

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564566795031, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=2, ext=[AuthorCompanyExt(id=1319955564575183640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China), AuthorCompanyExt(id=1319955564583572249, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学先进制造学院,福建 泉州 362000)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955565455987504, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955565539873588, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565455987504, language=EN, stringName=Zhixian Lin, firstName=Zhixian, middleName=null, lastName=Lin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, 3, address=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China
2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955565598593845, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565455987504, language=CN, stringName=林志贤, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 2, 3, address=1福州大学物理与信息工程学院,福建 福州 350116
2福州大学先进制造学院,福建 泉州 362000
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564478714644, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=1, ext=[AuthorCompanyExt(id=1319955564487103253, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564491297558, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学物理与信息工程学院,福建 福州 350116)]), AuthorCompany(id=1319955564566795031, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=2, ext=[AuthorCompanyExt(id=1319955564575183640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China), AuthorCompanyExt(id=1319955564583572249, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学先进制造学院,福建 泉州 362000)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955565661508407, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955565737005882, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565661508407, language=EN, stringName=Jianpu Lin, firstName=Jianpu, middleName=null, lastName=Lin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955565804114747, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565661508407, language=CN, stringName=林坚普, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2福州大学先进制造学院,福建 泉州 362000
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564566795031, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=2, ext=[AuthorCompanyExt(id=1319955564575183640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China), AuthorCompanyExt(id=1319955564583572249, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学先进制造学院,福建 泉州 362000)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955565904778045, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=5, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955566001247040, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565904778045, language=EN, stringName=Shanhong Lü, firstName=Shanhong, middleName=null, lastName=Lü, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955566085133121, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955565904778045, language=CN, stringName=吕珊红, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=2, 3, address=2福州大学先进制造学院,福建 泉州 362000
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564566795031, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=2, ext=[AuthorCompanyExt(id=1319955564575183640, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China), AuthorCompanyExt(id=1319955564583572249, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564566795031, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=2福州大学先进制造学院,福建 泉州 362000)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])]), Author(id=1319955566152241987, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, orderNo=6, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1319955566265488198, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955566152241987, language=EN, stringName=Tailiang Guo, firstName=Tailiang, middleName=null, lastName=Guo, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China
3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1319955566332597063, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, authorId=1319955566152241987, language=CN, stringName=郭太良, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=1, 3, address=1福州大学物理与信息工程学院,福建 福州 350116
3中国福建光电信息科学与技术创新实验室,福建 福州 350116, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1319955564478714644, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=1, ext=[AuthorCompanyExt(id=1319955564487103253, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564491297558, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564478714644, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=1福州大学物理与信息工程学院,福建 福州 350116)]), AuthorCompany(id=1319955564642292506, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, xref=3, ext=[AuthorCompanyExt(id=1319955564650681115, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China), AuthorCompanyExt(id=1319955564659069724, tenantId=1146029695717560320, journalId=1278651732997652489, articleId=1319955554672431797, companyId=1319955564642292506, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=3中国福建光电信息科学与技术创新实验室,福建 福州 350116)])])]
希治远1, 3 , 梅婷1, 3, 林珊玲2, 3, *, * , 林志贤1, 2, 3, 林坚普2, 3, 吕珊红2, 3, 郭太良1, 3
作者信息
  • 1福州大学物理与信息工程学院,福建 福州 350116
  • 2福州大学先进制造学院,福建 泉州 362000
  • 3中国福建光电信息科学与技术创新实验室,福建 福州 350116
通讯作者:
林珊玲,
作者简介:

作者贡献

希治远:研究设计、论文初稿撰写、算法实现与验证;梅婷:参与实验设计与结果分析;林珊玲:负责研究方向把控、论文结构审阅与最终定稿;林志贤,林坚普,吕珊红:协助完成基准对比实验;郭太良:提供实验理论指导。

希治远(2000-),男,硕士研究生,主要研究方向为工业缺陷检测。E-mail:

林珊玲(1991-),女,福建泉州人,博士,讲师,硕士研究生导师,福州大学先进制造学院电子信息系教师,主要从事主要研究新型显示技术、平板显示器驱动系统图像处理技术等。E-mail:

Printed circuit board defect detection based on LightRT-DETR algorithm
Zhiyuan Xi1, 3 , Ting Mei1, 3, Shanling Lin2, 3 , Zhixian Lin1, 2, 3, Jianpu Lin2, 3, Shanhong Lü2, 3, Tailiang Guo1, 3
Affiliations
  • 1School of Physics and Information Engineering, Fuzhou University, Fuzhou, Fujian 350116, China
  • 2School of Advanced Manufacturing, Fuzhou University, Quanzhou, Fujian 362000, China
  • 3Fujian Optoelectronic Information Science and Technology Innovation Laboratory, Fuzhou, Fujian 350116, China
  • Corresponding Author:
出版时间: 2026-02-25 doi: 10.12086/oee.2026.250117
文章导航
收藏切换

针对工业生产中印刷电路板(printed circuit board, PCB)缺陷检测精度和实时性不高的问题,本文提出了一种基于LightRT-DETR的PCB缺陷检测算法。首先,设计了一种高效的主干网络EfficientFastGLU-18,该网络通过引入部分卷积(partial convolution, PConv) 和卷积门控线性单元(convolutional gated linear unit, convolutional GLU) 机制来降低ResNet-18架构的计算复杂度和参数量,同时保持较好的特征提取能力。其次,采用级联分组注意力机制替代了RT-DETR中的多头注意力机制,通过特征分割与级联,增强输入特征的多样性,并降低不必要的计算开销。最后,提出了一种高效选择性特征金字塔网络(efficient selective feature pyramid network, ES-FPN),其主要由特征选择模块和串联RetBlockC3的特征融合模块构成,增强对细微缺陷的检测能力的同时减少计算冗余,从而实现更高效的实时缺陷检测。实验结果表明,本文改进的模型的mAP50达到了98.3%,mAP50-95值较基准模型RT-DETR提高了11.1%,GFLOPs下降了34.7%,Params降低了35.3%,FPS提升了约5.1%。这些优化使得LightRT-DETR模型在移动设备或嵌入式系统中更加适用。

缺陷检测  /  印刷电路板  /  主干网络  /  特征金字塔网络  /  计算冗余
Objective

The manufacturing of printed circuit boards (PCBs) constitutes the foundational infrastructure of the contemporary electronics industry. As the sector evolves in response to consumer demands for increasingly miniaturized, sophisticated, and high-performance devices, the physical characteristics of PCBs have undergone significant transformation. Modern boards are characterized by high-density interconnects, multi-layered architectures, and microscopic component placement. In this context, the role of quality control has shifted from a supplementary process to a critical determinant of product viability and yield. Traditionally, defect detection has relied on manual inspection or standard automated optical inspection (AOI) systems. However, these conventional methods are increasingly challenged by the escalating complexity of modern PCB designs. Manual inspection is labor-intensive, subjective, and prone to fatigue-induced errors, rendering it unsuitable for mass production. Similarly, traditional AOI systems based on pattern matching or simple image processing often struggle with the subtle variations and complex backgrounds found in high-density PCBs. Consequently, there is a growing necessity for more intelligent and adaptive inspection solutions capable of discerning defects with high precision. Current state-of-the-art detection methods, particularly those leveraging deep learning paradigms, encounter a distinct set of challenges when applied to the specific constraints of industrial manufacturing environments. The first challenge is the significant computational burden associated with high-precision models. Accurately detecting minute defects typically necessitates high-resolution imagery; however, processing such data with deep convolutional neural networks (CNNs) or complex Transformer models consumes substantial computational resources and memory. This computational demand often translates to prohibitive hardware costs for widespread deployment on factory floors. The second challenge is scale variation, which presents a persistent hurdle. PCB defects manifest in widely varying sizes, from tiny pinholes barely visible to the naked eye to larger structural errors like open circuits. A detection model must possess the multi-scale capability to identify these diverse anomalies with uniform accuracy. The third challenge is the imperative for real-time performance. In high-velocity production lines, inspection algorithms must operate at speeds commensurate with manufacturing throughput to avoid creating bottlenecks. Unfortunately, many existing high-accuracy models are computationally too intensive for real-time inference, forcing manufacturers to make an undesirable trade-off between inspection speed and detection precision. These limitations highlight the need for an optimized architecture that balances resource efficiency with robust detection capabilities.

Methods

To address these pressing challenges effectively, this study proposes LightRT-DETR, a lightweight real-time detection transformer specifically optimized for PCB defect detection. Building upon the architectural principles of the real-time detection Transformer (RT-DETR), our approach introduces a series of strategic modifications designed to mitigate computational costs significantly while maintaining, and in some aspects enhancing, detection accuracy. The design of LightRT-DETR is characterized by three primary technical innovations: the implementation of the EfficientFastGLU-18 backbone, the integration of a cascaded group attention (CGA) mechanism, and the development of an efficient selective feature pyramid network (ES-FPN). These components work in concert to refine the model's ability to extract relevant features from complex backgrounds without incurring the heavy computational penalties associated with standard transformer architectures. The overarching goal is to create a model that is both agile enough for edge deployment and accurate enough to meet rigorous quality standards. To resolve the inefficiency of traditional backbones like ResNet-18 when deployed on edge computing devices, we propose the integration of the EfficientFastGLU-18 backbone. This novel structure synthesizes partial convolution (PConv) with convolutional gated linear units (GLU). PConv operates by applying convolution to only a subset of input channels while leaving the remainder untouched for the next layer. This strategy effectively reduces memory access and floating-point operations (FLOPs) without compromising the flow of information. By coupling this efficient convolution strategy with GLU, which provides robust non-linear transformation capabilities, the backbone maintains strong feature extraction abilities. This is particularly crucial for identifying defects against the complex, texture-rich backgrounds of PCBs, where differentiating between a valid trace and a defect requires discerning subtle feature variations. Furthermore, to address the computational expense of standard multi-head attention mechanisms, we introduce a cascaded group attention (CGA) mechanism. Instead of processing all feature channels simultaneously, CGA splits feature maps into distinct groups and processes them through a cascaded structure. This approach allows the network to refine features progressively, as the output of one group's attention computation can inform the next, creating a hierarchical refinement process. This mechanism appears particularly beneficial for defects possessing intricate or fine-grained details, such as "mouse bites" or "spurs", which are frequently overlooked by coarser, global attention mechanisms. Finally, we tackle multi-scale defect detection through the efficient selective feature pyramid network (ES-FPN). This redesigned module employs selective feature integration and incorporates RetBlockC3 modules enhanced with manhattan self-attention (MaSA). Unlike standard self-attention, MaSA is specifically designed to capture long-range dependencies in horizontal and vertical directions. Given that PCB layouts consist largely of orthogonal lines, this domain-specific adaptation enables more effective recognition of defects across widely varying scales by aligning the attention mechanism with the geometric priors of the PCB domain.

Results and Discussions

The efficacy of LightRT-DETR was validated through extensive experiments on a public PCB defect dataset, revealing a significant improvement in both efficiency and accuracy compared to current state-of-the-art models, including the widely utilized YOLO series. In terms of quantitative performance, LightRT-DETR achieved a mean average precision (mAP50) of 98.3% and a stricter metric, mAP50-95, of 52.9%. When benchmarked against the advanced YOLOv9e model, our approach demonstrated an improvement of 3.0% in mAP50 and 1.8% in mAP50-95. These metrics suggest that the model is not only capable of detecting defects but also does so with high localization accuracy, minimizing false positives and false negatives—a critical requirement for industrial quality control. The experimental setup ensured fair comparisons by utilizing consistent hyperparameters and training environments, further reinforcing the validity of the observed performance gains.

Conclusions

Perhaps the most distinct advantage of LightRT-DETR lies in its computational efficiency. The model requires approximately 37.2 GFLOPs, representing an 80.3% reduction in computational cost compared to the baseline RT-DETR model. Similarly, the parameter count is reduced to 12.8 million, a 77.7% decrease. This substantial reduction has practical implications: it facilitates deployment on cost-effective hardware with limited computing power or allows the system to operate at significantly higher frame rates on standard hardware.

defect detection  /  printed circuit board  /  backbone network  /  feature pyramid network  /  computational redundancy
希治远, 梅婷, 林珊玲, 林志贤, 林坚普, 吕珊红, 郭太良. 基于LightRT-DETR算法的印刷电路板缺陷检测. 光电工程, 2026 , 53 (2) : 250117 -1 . DOI: 10.12086/oee.2026.250117
Zhiyuan Xi, Ting Mei, Shanling Lin, Zhixian Lin, Jianpu Lin, Shanhong Lü, Tailiang Guo. Printed circuit board defect detection based on LightRT-DETR algorithm[J]. Opto-Electronic Engineering, 2026 , 53 (2) : 250117 -1 . DOI: 10.12086/oee.2026.250117
印刷电路板(PCB)制造行业在电子工业的版图中占据着举足轻重的地位。然而,在PCB生产过程中,缺陷检测环节仍然面临着巨大挑战,传统的人工检测效率低下、准确性不足,难以满足现代电子制造业对高质量和高效率的严格要求。此外,像短路、开路、鼠咬等缺陷的检测会引起视觉疲劳,容易导致误分类[1]。为了克服这些问题,深度学习算法在PCB缺陷检测领域展现出巨大潜力。
目前,PCB缺陷检测的深度学习算法主要分为两大类:基于传统深度卷积神经网络(convolutional neural networks, CNN)的目标检测算法和基于Transformer架构的目标检测算法。基于CNN的缺陷检测算法又可细分为两类,单阶段算法与两阶段算法。单阶段算法主要包含YOLO[2]系列、RetinaNet[3]和单点多盒检测SSD[4]系列算法。两阶段算法主要包括R-CNN[5]、Fast R-CNN[6]、Faster R-CNN[7]和Mask R-CNN[8]。近年来,研究者们对这些算法进行了多方面的改进,Lian等[9]利用mask R-CNN进行缺陷检测,并在全连接CNN中添加几何注意力引导的mask分支,以提高模型的效率。Li等[10]将Faster R-CNN和YOLOv2融合,提高了PCB双列直插式组件的检测效率并降低了误检测率。Xin等[11]针对PCB的缺陷检测,提出了一种增强的YOLOv4模型。该方法引入马赛克数据增强策略和Mish激活函数,提高了小目标检测能力。Xie等[12]使用多阶段残差混合注意力模块MRHAM改进特征学习,并通过K-means++聚类优化锚框,显著提高了检测准确性。Wu等[13]提出GSC YOLOv5模型,通过Ghost Conv和Ghost Bottleneck实现轻量化,并引入双重注意力机制提高检测精度和效率。Wang等[14]提出基于轻量级深度学习的缺陷检测网络,使用CSPDarkNet作为YOLOX主干网络,并融入协调注意力机制,提高了小型PCB缺陷检测能力。陶志勇等[15]提出了 CViT-Net 网络,通过融合局部与全局特征提升缺陷检测性能,并设计轻量与标准两种结构以兼顾精度与效率。
基于Transformer的目标检测算法在处理全局信息、捕捉细节特征和建立序列关系方面展现出独特优势,相较于传统的CNN方法更具竞争力。基于此,Carion等[16]提出了一种完全端到端的检测器DETR,采用集合预测方法,巧妙地规避了非极大值抑制 (non-maximum suppression, NMS) 后处理步骤的需求。该模型运用匈牙利算法来实现对象分类和回归,展现出优秀的性能。Zhao等[17]提出了RT-DETR模型,RT-DETR使用一个高效的混合编码器取代了常规Transformer编码器。RT-DETR分离处理不同尺度特征的内部交互和跨尺度融合,这一策略显著提升了模型的推理速度。此外,RT-DETR采用了基于不确定性最小化的查询选择机制,这种方法通过精确优化不确定性,为解码器提供高质量的初始查询,从而提高了整体检测精度。
但是,现阶段RT-DETR也有些许不足之处,针对这些不足本文将对其进行优化。1) ResNet[18]作为RT-DETR的主干网络结构需要大量的计算资源和存储资源,尤其是在处理高分辨率图像时,这使得它在资源受限的环境中应用受到限制。所以本文设计出一种新的主干网络EfficientFastGLU-18优化其带来的计算负担。2) RT-DETR中的基于注意力的内部尺度特征交互AIFI采用了多头注意力机制MHA[19],需要计算多个注意力头,并且多个注意力头可能会捕捉到相似的信息,导致参数冗余,本文将采用级联分组注意力CGA[20],通过级联和分组的方式,提高了模型的计算效率。并且对于漏孔和鼠咬等需要精细局部特征的缺陷,CGA的逐层特征提取能力可以更准确地捕捉细微的形状变化和边缘特征。3) RT-DETR中的路径聚合网络(path aggregation network, PANet) 只能完全融合相邻层级的特征。这限制了模型对多样化缺陷的识别能力。同时,在融合高低层级特征时,PANet无法专注于特定的语义信息。这对检测如杂刺和杂散等细微缺陷尤为重要,因为这些缺陷可能表现为微小的异常纹理或颜色变化。此外,双向金字塔结构中存在大量路径和间接信息交互,显著增加了模型的计算量。基于此本文提出了ES-FPN网络解决上述问题。结合上述改进策略,本文设计了LightRT-DETR模型,这是一种适合执行PCB缺陷检测任务的轻量级实时目标检测模型。
RT-DETR是一种基于Transformer架构的实时端到端目标检测器,本文选择了RT-DETR-R18作为基础模型,该模型在检测精度和模型复杂度之间取得了良好平衡。如图1所示,RT-DETR-R18的网络结构主要包括ResNet18主干网络用于特征提取,高效混合编码器(包含AIFI模块和CCFM)负责多尺度特征转换,以及解码器部分。
本文从特征提取、Transformer结构、多尺度特征转换等方向出发,对RT-DETR-R18模型进行模块与结构改进,提出LightRT-DETR新方法,如图2所示。LightRT-DET模型由主干网络、混合编码器和带辅助预测头的变压器解码器组成。利用轻量级的主干网络输出特征作为编码器的输入。混合编码器通过基于CGA的尺度内交互机制与一种高效选择性特征金字塔网络将多尺度特征转换为一系列图像特征。随后使用IoU感知从编码器输出序列中选择固定数量的图像特征作为解码器的初始对象查询。最后,使用带有辅助预测头的解码器迭代优化对象查询以生成方框和置信度分数。
在现代计算机视觉任务中,神经网络的计算效率和参数数量是至关重要的性能指标。针对RT-DETR模型在处理高分辨率图像时资源消耗高的问题,本文首先对ResNet-18中的BasicBlock进行优化,引入FasterBlock来替代传统的卷积块。随后,为了进一步提高特征提取能力,本文引入了Convolutional GLU,对FasterBlock进行二次改进。这种改进后的模块被命名为EfficientFastGLU,如图3所示,其最终实现了在保持较高精度情况下提升计算效率和减少参数量的目标。
EfficientFastGLU中的核心之一是部分卷积。如图4所示,可以同时减少计算量和内存访问,更高效地提取空间特征。PConv利用特征图的冗余性,选择性地仅对部分输入通道(cp)进行卷积操作,而其余通道则使用恒等映射。假设输入和输出特征图的通道数量相同,这种方法不会丧失一般性。与传统的卷积操作对所有通道(c)进行卷积相比,PConv能够显著减少浮点运算次数,有效降低计算复杂度和内存访问量,如式(1)所示。
$ {r}_{{\mathrm{F}}}=\dfrac{h\times w\times {k}^{2}}{h\times w\times {k}^{2}\times {c}^{2}}\times \frac{{c}^{2}}{{r}^{2}}\;, $
式中:$ {r}_{{\mathrm{F}}} $为PConv的浮点运算次数与普通卷积浮点运算次数的比值;h、w分别为输入特征图的高和宽;k为卷积核大小;r为通道缩减因子,其取值为0.25,对于该取值安排,在第4节中进行了详细展示说明。由此可见,PConv的计算量仅为普通卷积的1/16,内存访问数量仅仅为普通卷积的1/4。在主干特征提取网络中引入PConv算子能够显著降低计算量和内存访问量,从而使骨干网络轻量化并提高模型的推理速度。
此外,EfficientFastGLU具有强大的特征提取能力是因为其引入了Convolutional GLU。这是对传统压缩-激励(squeeze-and-excitation, SE) [21]机制的改进。如图5所示,Convolutional GLU作为一种创新的通道混合器,克服了SE机制中全局平均池化导致的局限性。它由两个线性变换组成,一个通过非线性激活函数作为门控信号,另一个作为值分支。这使得每个token都能基于其邻近特征获得定制的门控信号,实现更精细和灵活的通道注意力调整。
EfficientFastGLU-18网络通过结合FasterBlock和Convolutional GLU,在保持与卷积前馈网络(convolutional feed-forward network, ConvFFN)相似参数量的前提下,显著降低了计算复杂度。这种设计不仅确保了模型的高效性和训练稳定性,还在ResNet-18的基础上实现了性能的显著提升。
本文使用了级联分组注意力机制CGA来代替AIFI中的多头注意力机制(multi-head attention, MHA),以此来提高模型性能和降低冗余参数。这种注意力机制可以表示为
$ {\tilde {X}}_{ij}=Attn\left({X}_{ij}{W}_{ij}^{{\boldsymbol{Q}}}\text{,}{X}_{ij}{W}_{ij}^{{\boldsymbol{K}}}\text{,}{X}_{ij}{W}_{ij}^{{\boldsymbol{V}}}\right) \;, $
$ {\tilde {X}}_{i+1}=Concat{\left[{\tilde {X}}_{ij}\right]}_{j}=_{1:h}{W}_{i}^{{\boldsymbol{P}}}\;, $
式中:Q是查询向量,代表当前位置想要查询的信息。K是键向量,代表所有位置的信息索引,用于和 Q 计算相似度,得到注意力权重。V是值向量,代表所有位置的实际内容,注意力权重会作用在V上,得到最终的加权输出。P是投影权重矩阵,用于对多头注意力拼接后的特征做线性变换,完成特征融合与维度映射。i代表网络层级索引,j代表多头注意力中的注意力头索引。第j个头计算Xij的自注意力,Xij是输入特征Xi的第j个分割部分,即Xi =[Xi1, Xi2, ···, Xih],且1≤j≤h。h是头的总数,$ W_{ij}^{\boldsymbol{Q}}、W_{ij}^{\boldsymbol{K}}、W_{ij}^{\boldsymbol{V}} $是将输入特征分割映射到不同子空间的投影层,而$ {W}_{i}^{P} $是一个线性层,将连接后的输出特征投影回与输入一致的维度。
CGA机制通过避免传统方法中对完整特征的全面依赖,转而采用特征分割与级联分组策略。具体来说,CGA机制首先将输入特征分成多组,每组包含若干个通道。然后,将不同的注意力头通过级联连接起来,形成级联注意力结构,如图6所示。在该结构中,每个注意力头只关注一组输入特征,并将注意力计算的输出传递给下一个注意力头。最后,级联结构的输出特征通过跨层连接与其他层的特征相连,增强特征表达。
$ {X}_{ij}'={X}_{ij}+{\tilde {X}}_{i\left(j-1\right)}\text{,}1 \lt j \lt h\;, $
式中:$ {X}_{ij}' $是第j个输入分割$ {X}_{ij} $和通过式(2)计算的第(j-1)个头输出$ {\tilde{X}}_{i\left(j-1\right)} $的相加结果。在计算自注意力时,它取代$ {X}_{ij} $作为第j个头的新输入特征。此外,在q投影之后应用了另一个token交互层,使自注意力能够共同捕捉局部和全局关系,进一步增强特征表示。CGA机制通过使用分组、级联、跨层连接等方式,可以增强输入特征的多样性,减少计算冗余,提高模型性能和计算效率。
RT-DETR模型中,CCFM使用了PANet来实现多尺度特征的融合。虽然这种方法通过自顶向下和自底向上的双向金字塔结构来增强特征的交互和定位能力,但存在以下三个主要问题:1) PANet只能完全融合相邻层级的特征。对于不相邻的层级,特征信息需要通过递归的方式间接获取。例如,漏孔和鼠咬可能需要更细粒度的特征,而开路和短路可能需要更大范围的上下文信息。仅依赖相邻层级的特征融合可能会限制模型对这些多样化缺陷的识别能力。2) 在融合高低层级特征时,PANet无法专注于特定的语义信息。这对检测如杂铜和杂散等细微缺陷尤为重要,因为这些缺陷可能表现为微小的异常纹理或颜色变化,需要模型能够捕捉非常细致的特征。3) 双向金字塔结构中存在大量路径和间接信息交互,显著增加了模型的计算量。在PCB生产线上进行实时缺陷检测时,检测速度直接影响生产效率,因此需要一个更高效的特征融合机制。为解决上述问题,本文提出一种高效选择性特征金字塔网络,其网络结构图如图7所示。其中,S3、S4、S5分别表示主干网络第3、4、5阶段输出的特征图,N3、N4、N5分别表示经过初步特征选择后的多尺度特征图。ES-FPN架构主要是由特征选择模块与串联RetBlockC3的特征融合模块构成。
通道注意力机制(channel attention, CA)模块和维度匹配(dimension matching, DM)模块在特征选择过程中协同发挥着关键作用。具体而言,CA模块,如图8所示,首先对输入特征图$ f_{\mathrm{in}}\in\mathbf{R}^{C\times H\times W} $进行处理,其中C表示通道数,H和W分别表示特征图的高度和宽度。该特征图分别经过全局平均池化和全局最大池化操作,获得两组全局描述信息,随后将两组池化特征进行融合,并通过一个Sigmoid激活函数生成每个通道的权重$ f_{\mathrm{CA}}\in\mathbf{R}^{C\times1\times1} $。这些权重与原始特征图逐通道相乘,实现对重要通道的增强和无效通道的抑制,从而有效提取每个通道中的关键信息。接下来,经过CA模块加权的特征图输入到DM模块,DM模块通过1×1卷积、上采样或下采样等操作对特征的空间分辨率和通道数进行调整,使不同尺度的特征在空间和通道维度上实现一致性匹配。通过这种先通道加权、后维度对齐的融合逻辑,特征选择模块能够充分挖掘多尺度特征的有效信息,并为后续的特征融合和检测任务提供高质量的特征基础。
特征融合模块通过选择性特征融合机制,使用高级特征作为权重来过滤低尺度特征中的重要信息。高级特征经过转置卷积和双线性插值操作进行尺度调整来匹配低尺度特征的尺寸,然后利用高级特征作为注意力权重,筛选出低尺度特征中有用的信息。并且通过将最高层特征N5不经处理直接输入解码器,最大程度地保留了其内部蕴含的丰富语义信息。为解决特征融合过程中的计算负担问题,本文从文献[22]中汲取灵感,串联了RetBlockC3模块,如图9所示,其中C1/2表示输入特征通道数C1的一半,C2表示模块输出的特征通道数。这一策略确保了筛选后的重要语义信息能够有效融合,同时提高计算效率。RetBlockC3主要是利用曼哈顿自注意力机制MaSA,减少了计算复杂度,直接降低了处理大规模数据集时的内存消耗。
MaSA将垂直方向上的注意力分数(经过值矩阵V加权后)进行转置,并与水平方向上的注意力分数相乘,得到最终的注意力输出MaSA(X)。这种分解和重组的方式,既保留了自注意力机制的全局感知能力,又通过分解降低了计算成本。
$ {\gamma }_{i}=\log\left(1-{2}^{-\left(2+\tfrac{4\cdot i}{num\_heads}\right)}\right) \;, $
$ {Attn}_{\rm{H}}=S oftmax({{\boldsymbol{Q}}}_{\rm{H}}{{\boldsymbol{K}}}_{\rm{H}}^{{\mathrm{T}}})\odot {{\boldsymbol{D}}}^{{\mathrm{H}}}\;, $
$ {Attn}_{\rm{W}}=S oftmax({{\boldsymbol{Q}}}_{\rm{W}}{{\boldsymbol{K}}}_{\rm{W}}^{{\mathrm{T}}})\odot {{\boldsymbol{D}}}^{\rm{W}}\;, $
$ MaS A\left({\boldsymbol{X}}\right)={Attn}_{\rm{H}}{({Attn}_{\rm{W}}{\boldsymbol{V}})}^{{\mathrm{T}}}\;, $
式中:$ {{\boldsymbol{D}}}_{nm}^{{\mathrm{H}}}={{\gamma }_{i}}^{|{y}_{n}-{y}_{m}|} $,$ {{\boldsymbol{D}}}_{nm}^{\rm{W}}={{\gamma }_{i}}^{\left|{x}_{n}-{x}_{m}\right|} $,其中n、m为像素点的索引;γ为衰减因子;i为注意力头的索引,i=0, 1, ···, num_heads−1,由于RT-DETR-R18共有8个注意力头,所以num_heads为8;$ {{\boldsymbol{Q}}}_{\rm{H}} $、$ {{\boldsymbol{K}}}_{\rm{H}} $分别为水平方向的查询和键;$ {{\boldsymbol{Q}}}_{\rm{W}} $、$ {{\boldsymbol{K}}}_{\rm{W}} $分别为垂直方向的查询和键;V为值矩阵;$ {{\boldsymbol{D}}}^{{\mathrm{H}}} $、$ {{\boldsymbol{D}}}^{\rm{W}} $分别为水平方向和垂直方向的位置衰减矩阵。
本文使用了北京大学智能机器人开放实验室发布的开源数据集,数据集中共有693张图片涵盖了6种PCB缺陷类型,包括漏孔、鼠咬、开路、短路、杂铜和杂散,每张图片的分辨率为2777 pixel×2138 pixel。每张图片包含3~6个缺陷。训练集、验证集和测试集划分比为7:1:2。具体数据集信息如表1所示。
实验平台的硬件环境为 GPU为RTX 3090,显存为24 GB,CPU为Intel Xeon Platinum 8362,内存为45 GB,操作系统及软件环境为Ubuntu 18.04、Cuda 11.1、Python 3.8、PyTorch 1.9.0。送入模型训练的图片大小均被设置为 640×640,batch_size 设置为 4,epoch设置为200,采用自适应权重衰减优化器AdamW,初始学习率为 0.0001,lrf设置为1.0,优化器权重衰减系数和动量系数分别设置为0.0001和0.9,颜色空间增强方面,hsv_h设置为0.015,hsv_s设置为0.7,hsv_v设置为0.4,几何变换增强方面,translate设置为0.1,fliplr设置为0.5,scale设置为0.5。
为了分析模型对PCB板的缺陷检测效果的检测效果,选择常用的精确率(precision, P)、召回率(recall, R)、平均精度均值(mAP)作为衡量算法性能的主要指标。选择模型参数量(Params)、计算量(GFLOPs)、检测速度(FPS)等参数作为衡量模型推理性能的指标。
$ R=\frac{TP}{TP+FN}\;, $
$ P=\frac{TP}{TP+FP}\;, $
$ AP={\int }_{0}^{1}P(R) {\mathrm{d}}R\;, $
$ mAP=\dfrac{\displaystyle{\sum }_{j=1}^{C}{AP}_{j}}{C}\;. $
其中:TP是真正例,即模型正确识别的正样本数量。FN是假反例,即模型错误判定为负样本的正样本数量。FP是假正例,即模型错误判定为正样本的负样本数量。
为了验证主干网络改进的有效性,本文在RT-DETR的基础上,选择了不同的轻量化方案作为对比。对比实验在验证集上的实验结果如表2所示。
由表2可知,EfficientFastGLU-18在目标检测任务中具有较好性能。与其他方案相比,EfficientFastGLU-18不仅降低了计算复杂度,GFLOPs从57.0降至46.2,参数量也从19.8 M减少到15.4 M,还保持了较高的推理速度,达到57.2 f/s,最关键的是该模型实现了最高的检测精度,其mAP50达到96.9%。这种全面的性能提升源于其创新的网络设计:部分卷积显著提高了计算效率,Convolutional GLU增强了特征提取能力,而整体网络结构的优化则在效率和精度之间取得了良好平衡。相比之下,其他模型如 EfficientViT和RepViT虽然在轻量化方面表现出色,但检测精度分别降至90.0%和88.4%,可能是过度追求效率而牺牲了特征提取能力。Fasternet和MobileNet_v4虽然在FPS方面领先,分别达到64.9和64.0 f/s,但mAP50相对较低,分别为94.4%和91.5%,表明它们可能在复杂特征提取方面存在不足。EfficientFormer_v2虽然保持了较高的mAP50,但FPS明显降低,反映了在特征提取效率和精度平衡上的挑战。这些对比凸显了EfficientFastGLU-18在目标检测任务中实现高效率、高精度平衡的良好表现。
本文在EfficientFastGLU-18主干网络中引入了部分卷积,部分卷积Pconv中的一个重要参数就是通道缩减因子r。r过小会导致信息损失,影响特征表达能力,r过大会引入冗余计算,降低模型效率,对此本文对其进行了系统验证。结果如表3所示,通道缩减因子r=0.250是精度与效率的最佳平衡点。实验数据表明,相比r=0.125,r=0.250使mAP50提高了1.2%,计算量仅微增1.5%;相比r=0.500,r=0.250保持相同的检测精度,同时计算量减少5.7%,推理速度提高7.1%。这一选择不仅在理论上平衡了信息保留与计算效率,也在实践中证明其能够有效捕获关键特征信息,同时避免冗余计算。
由于在EfficientFastGLU-18中引入了PConv,所以可能会导致特征图表达能力减弱,并且引入梯度非连续性或退化学习问题,由此本文对该问题进行了验证。验证结果如图10与图11所示,EfficientFastGLU-18结构在特征表达能力、梯度连续性和学习稳定性方面均展现出显著优势。首先,在特征表达能力方面,该结构不仅未降低模型表现,反而显著提升了特征谱能量。图11(b)显示所有层的相似度都在0.4~0.5之间,说明EfficientFastGLU-18模型在这些层的特征分布与原始模型有一定差异,但仍有较高的相关性;图11(c) 显示EfficientFastGLU-18模型在各层的特征谱能量均高于原始模型,尤其在第4层提升约2倍。同时,如图10(b)、图11(a)显示,EfficientFastGLU-18模型在mAP50性能指标上持续优于原始模型。其次,关于梯度非连续性问题,如图10(a)显示,EfficientFastGLU-18模型的损失曲线平滑下降,无异常波动,训练过程无震荡现象,EfficientFastGLU-18模型收敛更快。若存在梯度非连续性应表现为收敛不稳定或迟缓,但实际并未观察到该现象。同时,在退化学习问题方面,图10(b)显示PConv模型在训练初期即可达到较高性能,第30个epoch时mAP提升高达38.9%,并且在整个训练周期内各项指标持续领先,无性能退化迹象。EfficientFastGLU-18的结构创新在于通道分离和门控机制的结合,一方面通过部分卷积高效提取局部特征,另一方面保留原始信息并增强特征选择能力,从而在GFLOPs从57.0降至46.2、参数量从19.8 M减少到15.4 M的同时,保持了较高的推理速度(57.2 f/s)和最高(96.9%)的平均精度均值。综上,EfficientFastGLU-18是一种理论合理、实验有效的高效结构优化方法,在提升计算效率的同时,显著增强了模型性能和泛化能力。
结合表4数据可以看出,CGA Attention在目标检测任务中展现出显著的综合性能优势。与其他主流注意力机制相比,CGA实现了最高的检测精度,其mAP50达到97.7%。这一性能不仅明显优于基准模型Baseline的96.2%,也超越了Deformable Attention的95.2%,以及多种轻量化注意力机制。在计算复杂度方面,CGA的GFLOPs为57.0,与Baseline和Deformable Attention等方法基本一致,说明该机制并未引入额外的计算开销。模型参数量方面,CGA的参数规模为19.7 M,亦与其他方法保持在相同数量级。从推理速度来看,CGA达到了57.8 f/s,明显优于Baseline的56.4 f/s和Deformable Attention的55.0 f/s,同时接近Performer的58.7 f/s和Linformer的60.3 f/s。虽然部分轻量化机制在推理速度上稍具优势,但其精度普遍低于CGA,特别是Focused Linear Attention和HiLo Attention,在降低计算成本的同时,其检测精度也出现了明显下滑。综合而言,CGA Attention通过引入创新的上下文引导机制,在不增加显著计算负担的前提下,有效提升了特征表达能力与目标检测精度。
结合表5的数据可以看出,ES-FPN 在目标检测任务中展现出优异的综合性能。在检测精度方面,ES-FPN实现了97.4%的mAP50,显著优于基准特征融合结构PANet的96.2%,同时也超过了其他主流方法,如BIFPN的95.5%、HSFPN的96.4%以及SDFM的95.8%,充分体现了其在特征表达能力方面的显著优势。在计算效率方面,ES-FPN的计算复杂度为47.9 GFLOPs,在所有对比方法中最低,明显优于BIFPN的64.3和PSFM的71.2。这表明ES-FPN能够在保持高检测精度的同时,有效降低计算开销。此外,其参数量仅为17.4 M,为所有结构中最小的,进一步彰显该模型在轻量化设计方面的优势。ES-FPN 通过引入通道注意力机制,有效增强非相邻层级间的信息交互能力,提升对多样化缺陷的识别效果。同时,该结构能够突出关键语义通道,增强了对细微缺陷的检测能力。配合轻量化的RetBlockC3模块和MaSA注意力机制,ES-FPN有效降低计算负担,提升推理速度,兼顾检测精度与效率。
为了验证所提方法中各模块的有效性,本文进行了一系列消融实验,结果如表6所示。基准模型M1不包含任何提出的模块,而M2至M7分别添加了不同组合的模块。实验结果表明,各模块均对检测性能产生了积极贡献。在M2中单独引入EfficientFastGLU-18模块后,mAP50提升了0.7%,同时计算量减少18.9%,参数规模下降22.2%。这一结果充分证明该模块能够在提升模型精度的同时实现显著的轻量化优化。M3所引入的CGA Encoder 模块带来了1.5%的mAP50提升,表明增强模型对全局上下文信息的建模能力对于提高检测效果至关重要。M4中加入的ES-FPN模块在提升检测精度1.2%的同时,计算量减少了16.0%,体现出该结构在多尺度特征融合方面的高效性与实用性。在模块组合方面,M5联合使用EfficientFastGLU-18与ES-FPN,在检测精度和计算效率之间实现了良好平衡。M7则结合了CGA Encoder与ES-FPN,mAP50达到97.8%,为所有非完整模型中精度最高的组合,验证了其在特征增强与融合方面的互补优势。最终的完整模型M8整合了所有三个模块,实现了最佳的98.3%的mAP50,同时保持了低计算复杂度仅为37.2的GFLOPs和12.8 M的小模型尺寸,推理速度达到59.3 f/s。三大模块各有侧重,联合使用时能实现性能的全面提升,展现出明显的协同增益效应。
尽管前述消融实验系统性地评估了各改进模块对整体检测性能的提升效果,但在实际缺陷检测任务中,不同类型目标(如小目标与大目标)由于其特征分布和检测难度的差异,可能对改进策略的响应存在显著不同。为进一步验证各模块在不同目标尺寸下的适用性及优势,本文对各模块在小目标与大目标检测任务中的表现进行了细致分析。具体而言,依据小目标目前的通用定义将测试集中缺陷面积小于32 pixel×32 pixel的目标定义为小目标,大于该阈值的定义为大目标,并分别对两类目标进行检测性能评估,相关结果见表7。
从结果可以看出,所提出的三大改进模块在提升大、小目标检测性能方面均展现出积极效果,但各自优化机制侧重点不同。EfficientFastGLU-18模块所构建的模型M2通过引入轻量化深层注意力机制,有效提升了特征提取效率,尤其在小目标检测方面表现突出,其小目标mAP50-95由基础模型的47.5%提升至51.8%。与此同时,大目标检测精度也得到一定提升,显示出对多尺度特征的良好适应能力。CGA Encoder模块对应的模型M3进一步强化了局部空间建模与上下文理解能力,使大目标mAP50达到98.4%,小目标的mAP50-95提升至52.3%。该模块在增强全局感知和特征聚合方面的能力,显著提升了复杂目标结构的检测效果。ES-FPN模块所构建的模型M4通过跨层次特征融合,有效整合了浅层高分辨率信息与深层语义表达,提升了大目标检测精度至98.0%,小目标检测性能也达到97.2%,其mAP50-95提高至52.2%,表明其在多尺度融合方面具备良好效果。在此基础上,多个模块的组合进一步强化了模型性能。模型M7联合CGA Encoder与ES-FPN,在大目标的mAP50-95上达到最高值56.0%。最终模型M8整合了EfficientFastGLU-18、CGA Encoder与ES-FPN三大模块,实现了在大、小目标检测精度上的全面最优表现,大目标mAP50达到98.5%,小目标mAP50达到98.2%,小目标mAP50-95提升至52.4%。
上述实验结果表明,三大模块具有良好的互补性。EfficientFastGLU-18提升了不同尺度目标的快速识别能力,CGA Encoder加强了小目标边缘细节与大目标结构理解,ES-FPN有效平衡了小目标的精细定位与大目标的语义表达。最终集成模型在不同目标尺度下均实现最优或近最优性能,充分验证了所提方法的鲁棒性与广泛适应性。
为了验证本文所提算法的优越性,本文选择了几种主流的目标检测算法进行对比,包括YOLOv5l、YOLOv8m、YOLOv8l、YOLOv9m和YOLOv9e,同时也包括文献[23]提到的EP-RTDETR、文献[24]提到的基于Faster R-CNN的改进模型和文献[25]提到的LIDD-YOLO,同时还涵盖了WiseYOLOv7[26]、FPFM[27]、LDD-Net[28]、ExtendedFPN[29]和TD-Net[30]等近年来代表性方法。本文所提LightRT-DETR算法在多个关键指标上均表现出色。实验结果如表8所示。首先,在检测精度方面,EP-RTDETR缺乏对PCB缺陷检测任务的针对性设计,导致其mAP50比本文算法的提到的低1.6%,并且在模型规模上处于明显劣势。尽管WiseYOLOv7的mAP50-95高于本文算法,但是其计算量是本文算法的3.6倍,具有较高计算复杂度。本文算法的优异性能主要源于三个关键创新:首先,引入EfficientFastGLU-18主干网络,有效降低了高分辨率PCB图像处理的资源需求;其次,采用级联分组注意力替代多头注意力机制,显著提高了参数效率,特别适合捕捉细微缺陷特征;最后,提出ES-FPN网络优化特征融合,增强了对多样化PCB缺陷的识别能力。总体来看,虽然YOLOv系列、WiseYOLOv7等模型在通用目标检测任务中表现优异,但在PCB缺陷检测这一特定任务上,存在模型规模庞大、易过拟合或参数效率低等局限性。部分轻量级模型如LIDD-YOLO、LDD-Net、TD-Net等虽然模型小巧,但检测精度略显不足。
图12展示了RT-DETR与LightRT-DETR两种模型在训练过程中的性能对比,从图12(a)可以看出,两个模型的损失值均随训练轮数增加而迅速下降,且LightRT-DETR的损失下降速度更快,收敛速度优于RT-DETR。在约100个epoch后,两个模型的损失值趋于稳定,但LightRT-DETR的损失始终低于RT-DETR。图12(b)进一步显示,随着训练的进行,两个模型的mAP50均快速提升,但LightRT-DETR的提升速度明显快于RT-DETR,并且在整个训练周期内始终保持更高的mAP50,最终达到更优的检测精度。这一显著差异不仅体现了LightRT-DETR在训练过程中的持续优势,也表明其在最终模型精度和泛化能力方面更具竞争力。
如图13所示,根据LightRT-DETR和RT-DETR-R18在PCB缺陷检测中的表现,可以观察到LightRT-DETR在多个方面展现出明显优势。对于漏孔和开路类缺陷,虽然两种模型都能成功检测,但LightRT-DETR表现出更高的置信度。这一特性在实际应用中尤为重要,因为高置信度的检测结果可以减少人工复查的需求,提高整体检测效率。在处理短路、杂铜和杂散等类型的缺陷时,LightRT-DETR的优势更为明显。RT-DETR-R18在这些缺陷类型上存在一定程度的漏检情况,而改进后的LightRT-DETR则展现出全面而准确的检测能力,几乎不遗漏任何缺陷。这种全面的检测能力对于保证PCB生产质量至关重要。此外,对于鼠咬这类较为特殊和难以识别的缺陷,RT-DETR-R18表现出明显的不足,识别能力较弱。相比之下,经过改进的LightRT-DETR在这一问题上取得了显著进步,能够较好地识别和定位鼠咬缺陷。这一改进不仅展示了LightRT-DETR的技术优势,也凸显了其在处理复杂和多样化PCB缺陷方面的潜力。
图14展示了原始标注、RT-DETR-R18和LightRT-DETR在PCB缺陷检测中的热力图对比。分析结果表明,LightRT-DETR相较于RT-DETR-R18取得了显著进步。LightRT-DETR生成的热力图呈现出更加集中和明确的热点分布,特别是在PCB板的关键区域,如PCB开路缺陷处。这种改进表明LightRT-DETR具有更高的定位精度和更强的特征提取能力,能够更准确地识别潜在的缺陷位置。相比之下,RT-DETR-R18的热点分布相对分散,显示出在具体缺陷定位方面的不确定性。LightRT-DETR的这一优化不仅提高了检测的准确性,还降低了误检率,对PCB制造质量控制的提升具有重要意义。
本文针对PCB缺陷检测中面临的挑战,提出了基于RT-DETR的LightRT-DETR算法优化。通过设计高效主干网络EfficientFastGLU-18、引入级联分组注意力机制、提出高效选择性特征金字塔网络等一系列措施,显著提升了模型对各类PCB缺陷的检测能力和精度。实验结果表明,LightRT-DETR在检测精度、速度和计算效率方面均显示出显著优势。未来,将继续深化对模型性能的研究,进一步优化PCB缺陷检测技术,为推动智能制造技术的发展做出贡献

参考文献 引证文献
排序方式:
1
Lan Z, Hong Y, Li Y. An improved YOLOv3 method for PCB surface defect detection[C]//2021 IEEE International Conference on Power Electronics, Computer Applications (ICPECA), 2021: 1009–1015. https://doi.org/10.1109/ICPECA51329.2021.9362675.
2
Redmon J, Divvala S, Girshick R, et al. You only look once: unified, real-time object detection[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, 2016: 779–788. https://doi.org/10.1109/CVPR.2016.91.
3
Lin T Y, Goyal P, Girshick R, et al. Focal loss for dense object detection[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision, 2017: 2999–3007. https://doi.org/10.1109/ICCV.2017.324.
4
Liu W, Anguelov D, Erhan D, et al. SSD: single shot MultiBox detector[C]//14th European Conference on Computer Vision, 2016: 21–37. https://doi.org/10.1007/978-3-319-46448-0_2.
5
Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014: 580–587. https://doi.org/10.1109/CVPR.2014.81.
6
Girshick R. Fast R-CNN[C]//Proceedings of the 2015 IEEE International Conference on Computer Vision, 2015: 1440–1448. https://doi.org/10.1109/ICCV.2015.169.
7
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 29th International Conference on Neural Information Processing Systems, 2015: 91–99.
8
He K M, Gkioxari G, Dollár P, et al. Mask R-CNN[C]//Proceedings of the 2017 IEEE International Conference on Computer Vision, 2017: 2980–2988. https://doi.org/10.1109/ICCV.2017.322.
9
Lian J, Wang L T, Liu T Y, et al. Automatic visual inspection for printed circuit board via novel Mask R-CNN in smart city applications[J]. Sustain Energy Technol Assess, 2021, 44: 101032.
10
Li Y T, Kuo P, Guo J I. Automatic industry PCB board DIP process defect detection with deep ensemble method[C]//2020 IEEE 29th International Symposium on Industrial Electronics (ISIE), 2020: 453–459. https://doi.org/10.1109/ISIE45063.2020.9152533.
11
Xin H J, Chen Z B, Wang B Y. PCB electronic component defect detection method based on improved YOLOv4 algorithm[J]. J Phys Conf Ser, 2021, 1827(1): 012167.
12
Xie H F, Li Y, Li X C, et al. A method for surface defect detection of printed circuit board based on improved YOLOv4[C]//2021 IEEE 2nd International Conference on Big Data, Artificial Intelligence and Internet of Things Engineering (ICBAIE), 2021: 851–857. https://doi.org/10.1109/ICBAIE52039.2021.9390006.
13
Wu L G, Zhang L, Zhou Q. Printed circuit board quality detection method integrating lightweight network and dual attention mechanism[J]. IEEE Access, 2022, 10: 87617−87629.
14
Wang X, Gao J S, Hou B J, et al. A lightweight modified YOLOX network using coordinate attention mechanism for PCB surface defect detection[J]. IEEE Sens J, 2022, 22(21): 20910−20920.
15
陶志勇, 何燕, 林森, 等. 局部和全局特征融合的太阳能电池片表面缺陷检测[J]. 光电工程, 2024, 51(1): 230292.
Tao Z Y, He Y, Lin S, et al. Surface defect detection of solar cells using local and global feature fusion[J]. Opto-Electron Eng, 2024, 51(1): 230292.
16
Carion N, Massa F, Synnaeve G, et al. End-to-end object detection with transformers[C]//16th European Conference on Computer Vision, 2020: 213–229. https://doi.org/10.1007/978-3-030-58452-8_13.
17
Zhao Y, Lv W Y, Xu S L, et al. DETRs beat YOLOs on real-time object detection[C]//Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 16965–16974. https://doi.org/10.1109/CVPR52733.2024.01605.
18
He K M, Zhang X Y, Ren A Q, et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition, 2016: 770–778. https://doi.org/10.1109/CVPR.2016.90.
19
Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems, 2017: 6000–6010.
20
Liu X Y, Peng H W, Zheng N X, et al. EfficientViT: memory efficient vision transformer with cascaded group attention[C]//Proceedings of the 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023: 14420–14430. https://doi.org/10.1109/CVPR52729.2023.01386.
21
Hu J, Shen L, Sun G. Squeeze-and-excitation networks[C]//Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2018: 7132–7141. https://doi.org/10.1109/CVPR.2018.00745.
22
Fan Q H, Huang H B, Chen M R, et al. RMT: retentive networks meet vision transformers[C]//Proceedings of the 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024: 5641–5651. https://doi.org/10.1109/CVPR52733.2024.00539.
23
吴斌斌, 张礼华, 刘军伟, 等. 基于改进的EP-RTDETR小目标PCB表面缺陷检测[J]. 制造技术与机床, 2025, (3): 139−148.
Wu B B, Zhang L H, Liu J W, et al. Improved EP-RTDETR based surface defect detection on PCB[J]. Manuf Technol Mach Tool, 2025, (3): 139−148.
24
陈仁祥, 詹赞, 胡小林, 等. 基于多注意力Faster RCNN的噪声干扰下印刷电路板缺陷检测[J]. 仪器仪表学报, 2021, 42(12): 167−174.
Chen R X, Zhan Z, Hu X L, et al. Printed circuit board defect detection based on the multi-attentive faster RCNN under noise interference[J]. Chin J Sci Instrum, 2021, 42(12): 167−174.
25
Luo S, Xu Y P, Zhang C L, et al. LIDD-YOLO: a lightweight industrial defect detection network[J]. Meas Sci Technol, 2024, 36(1): 0161b5.
26
陈建豪, 徐洁, 汪志锋. 基于深度学习的PCB缺陷检测[J]. 现代电子技术, 2025, 48(8): 7−12.
Chen J H, Xu J, Wang Z F. PCB defect detection based on deep learning[J]. Mod Electron Tech, 2025, 48(8): 7−12.
27
Wang H D, Xie J, Xu X Y, et al. Few-shot PCB surface defect detection based on feature enhancement and multi-scale fusion[J]. IEEE Access, 2022, 10: 129911−129924.
28
Zhang L X, Chen J S, Chen J G, et al. LDD-Net: lightweight printed circuit board defect detection network fusing multi-scale features[J]. Eng Appl Artif Intell, 2024, 129: 107628.
29
Li C J, Qu Z, Wang S Y, et al. A method of defect detection for focal hard samples PCB based on extended FPN model[J]. IEEE Trans Compon Packaging Manuf Technol, 2022, 12(2): 217−227.
30
Shao R, Zhou M L, Li M, et al. TD-Net: tiny defect detection network for industrial products[J]. Complex Intell Syst, 2024, 10(3): 3943−3954.
2026年第53卷第2期
PDF下载
158
68
引用本文
BibTeX
文章信息
doi: 10.12086/oee.2026.250117
  • 接收时间:2025-04-19
  • 首发时间:2026-10-22
  • 出版时间:2026-02-25
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2025-04-19
  • 修回日期:2025-06-23
  • 录用日期:2025-06-23
基金
作者信息
    1福州大学物理与信息工程学院,福建 福州 350116
    2福州大学先进制造学院,福建 泉州 362000
    3中国福建光电信息科学与技术创新实验室,福建 福州 350116

通讯作者:

林珊玲,
参考文献
分享链接
https://castjournals.cast.org.cn/joweb/oee/CN/10.12086/oee.2026.250117
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数
科
Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)
属
Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏