Scholarlink

For electronic equipment to perform properly, a printed circuit board assembly (PCBA) quality is crucial. Significant research and development efforts were made to automate PCBA inspection for flaw detection, mostly using computer vision techniques, in order to meet the necessary quality standards. Despite these developments, different board and component sizes frequently affect how accurate these procedures are. Attempts to improve its accuracy, particularly for minute or minor flaws on a PCBA, frequently result in a trade-off of decreased real-time performance, which restricts its use in the manufacturing sector. In this study, we conduct a systematic comparison of four state‑of‑the‑art YOLO object‑detection models—YOLO v8, v9, v10, and v11—for printed circuit board (PCB) defect detection. A custom dataset of high‑resolution PCB images was manually annotated in nine classes (Capacitor, Diode, IC, Inductor, LED, Missing, Resistor, Sensor, Transistor); all images were standardized to 640 × 640 pixels and augmented with flips, crops, rotations, brightness adjustments, and blurs to enhance robustness. Among the medium variants, YOLO v10_m achieved the highest mAP@50 of 22.4% (precision 0.482, recall 0.242), while YOLO v8_m provided the fastest average inference time of 4.36 ms per image. These results illuminate the trade‑off between detection accuracy and speed, offering practical guidance for deploying YOLO models in real‑time PCB inspection pipelines.

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

Comparative Analysis of YOLO Model in Detecting PCB Defects with Associated Components

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jigar Sarda,
  • Hirva Patel,
  • Rohan Vaghela

摘要

For electronic equipment to perform properly, a printed circuit board assembly (PCBA) quality is crucial. Significant research and development efforts were made to automate PCBA inspection for flaw detection, mostly using computer vision techniques, in order to meet the necessary quality standards. Despite these developments, different board and component sizes frequently affect how accurate these procedures are. Attempts to improve its accuracy, particularly for minute or minor flaws on a PCBA, frequently result in a trade-off of decreased real-time performance, which restricts its use in the manufacturing sector. In this study, we conduct a systematic comparison of four state‑of‑the‑art YOLO object‑detection models—YOLO v8, v9, v10, and v11—for printed circuit board (PCB) defect detection. A custom dataset of high‑resolution PCB images was manually annotated in nine classes (Capacitor, Diode, IC, Inductor, LED, Missing, Resistor, Sensor, Transistor); all images were standardized to 640 × 640 pixels and augmented with flips, crops, rotations, brightness adjustments, and blurs to enhance robustness. Among the medium variants, YOLO v10_m achieved the highest mAP@50 of 22.4% (precision 0.482, recall 0.242), while YOLO v8_m provided the fastest average inference time of 4.36 ms per image. These results illuminate the trade‑off between detection accuracy and speed, offering practical guidance for deploying YOLO models in real‑time PCB inspection pipelines.