Scholarlink

<p>Printed Circuit Boards (PCBs) are the backbone of modern electronics, enabling the function and connectivity of countless everyday devices. Even minor defects in PCBs can compromise product quality and lead to failures. In recent years, Machine Learning (ML) and Deep Learning (DL) techniques have significantly advanced the detection of surface and solder joint defects in PCBs. To understand the progression, a Systematic Literature Review (SLR) was conducted using the PRISMA framework, analyzing 105 peer-reviewed articles published between 2000 and 2025. The review explored model performance, dataset characteristics, evaluation metrics, defect types addressed, and the practical advantages and limitations of various methods. Compared to other DL architectures, YOLO-based models are lightweight, enabling higher FPS during inference while maintaining high accuracy. Beyond technical comparisons, the review also analyzed regional research distribution and year-wise publication trends, offering insights into the geographic focus and temporal evolution of PCB defect detection research. The comprehensive review serves as a key resource for both researchers and engineers through in-depth perspectives on the development, evaluation, and optimization of object detection models in the context of PCB defect detection.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

PCB Defects: A Unified Survey of Trends, Detection Techniques, and Limitations through Systematic Literature Review

  • 期刊论文

  • 发表日期：

    2025年10月15日

查看全文
  • Shreevatsa Alawandi,
  • Kaushik Mallibhat,
  • Umer Kudachi,
  • Aishwarya Beedanal

摘要

Printed Circuit Boards (PCBs) are the backbone of modern electronics, enabling the function and connectivity of countless everyday devices. Even minor defects in PCBs can compromise product quality and lead to failures. In recent years, Machine Learning (ML) and Deep Learning (DL) techniques have significantly advanced the detection of surface and solder joint defects in PCBs. To understand the progression, a Systematic Literature Review (SLR) was conducted using the PRISMA framework, analyzing 105 peer-reviewed articles published between 2000 and 2025. The review explored model performance, dataset characteristics, evaluation metrics, defect types addressed, and the practical advantages and limitations of various methods. Compared to other DL architectures, YOLO-based models are lightweight, enabling higher FPS during inference while maintaining high accuracy. Beyond technical comparisons, the review also analyzed regional research distribution and year-wise publication trends, offering insights into the geographic focus and temporal evolution of PCB defect detection research. The comprehensive review serves as a key resource for both researchers and engineers through in-depth perspectives on the development, evaluation, and optimization of object detection models in the context of PCB defect detection.