Intelligent Detection of Passive and Active Components in PCB Imagery Using YOLOv12 for Automated Inspection Systems

Automated visual inspection of printed circuit boards (PCBs) is a critical task in modern electronics manufacturing, demanding high accuracy, robustness, and real-time performance. In this study, we present a comprehensive evaluation of YOLOv12, the latest and most advanced model in the YOLO object detection series, for the task of PCB component detection. Building on the strengths of its predecessors, YOLOv12 integrates architectural and optimization enhancements to improve detection precision, particularly for densely packed and visually ambiguous components such as capacitors and resistors. A custom dataset comprising annotated images of various PCB layouts, including Arduino and ESP-based development boards, was used for training and evaluation. The model demonstrated high precision and recall across most component classes, outperforming YOLOv5, YOLOv8, YOLOv10, and YOLO11 in both qualitative and quantitative metrics. Through detailed loss convergence analysis, class-wise performance curves, and confusion matrix evaluation, YOLOv12 proved its suitability for deployment in real-time automated inspection pipelines. The results highlight its potential to significantly enhance quality control processes in automated electronics manufacturing by reducing inspection time and human error, thus advancing the broader goals of intelligent automation in industry.