Scholarlink

<p>Printed circuit boards (PCBs) are essential components of electronic devices, and any minor defect can significantly affect their functionality in these devices. Identification of these minor defects is necessary to ensure the dependability of electronic equipments, however, current approaches are limited in terms of accurately identifying these defects. To overcome these limitations, we propose a defect detection approach based on the variants of You Only Look Once (YOLO), i.e., YOLOv8 and YOLOv9. The proposed approach is extensively evaluated on six classes of bare PCB images using a modified, publicly available dataset in terms accuracy, speed, and model complexity. Furthermore, we performed comparative analysis with different image resolutions where the YOLOv9 achieves the highest performance, with a mean Average Precision (mAP) of 98.4% at an IoU threshold of 0.5.</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

A YOLOv9: Deep Learning-based Framework Defect Detection Method for PCBs

  • 期刊论文

  • 发表日期：

    2025年8月1日

查看全文
  • Sana Nasri,
  • Nadeem Ahmad,
  • Qurat Ul Ain Aini,
  • Atif Qayyum,
  • Naeem Ul Islam

摘要

Printed circuit boards (PCBs) are essential components of electronic devices, and any minor defect can significantly affect their functionality in these devices. Identification of these minor defects is necessary to ensure the dependability of electronic equipments, however, current approaches are limited in terms of accurately identifying these defects. To overcome these limitations, we propose a defect detection approach based on the variants of You Only Look Once (YOLO), i.e., YOLOv8 and YOLOv9. The proposed approach is extensively evaluated on six classes of bare PCB images using a modified, publicly available dataset in terms accuracy, speed, and model complexity. Furthermore, we performed comparative analysis with different image resolutions where the YOLOv9 achieves the highest performance, with a mean Average Precision (mAP) of 98.4% at an IoU threshold of 0.5.