EVALUATION OF PRINTED CIRCUIT BOARD DEFECT DETECTION ALGORITHMS BASED ON FASTER R-CNN AND YOLOV8

Tóm tắt

Detecting printed circuit board defects is one of the critical tasks to ensure the quality of electronic devices, especially as printed circuit board sizes become increasingly compact, leading to higher demands for accuracy and speed in defect detection in industrial manufacturing. However, traditional inspection methods are time-consuming and inefficient for modern printed circuit boards. In recent years, deep learning techniques have demonstrated superior capabilities in detecting and classifying printed circuit board defects, providing a robust alternative to conventional methods. This paper presents an enhancement to the baseline model by incorporating modern techniques to analyze data in object detection tasks. By separately approaching the two models, Faster R-CNN and YOLOv8, we experimented with and compared their performance. Experimental results indicate that both models achieve promising performance; However, Faster R-CNN excels in accuracy, while YOLOv8 stands out for its speed.