Lightweight computer vision framework for defect detection in PCB manufacturing environments

Defect detection is the most critical factor in the manufacturing of Printed Circuit Boards (PCB) to ensure the reliability of the product and its performance in operation. Since PCBs are becoming more complex, traditional inspection methods that depend on manual or basic automated systems face many challenges, including high costs, extended times for inspection, and the inability to detect subtle defects. This has led to a growing need for more advanced, efficient, and accurate inspection techniques. To overcome the above difficulties, this research presents a lightweight computer vision framework for real-time PCB manufacturing defect detection. The method proposed here uses Intelligent Squirrel Search-driven Effective Generative Adversarial Networks (ISS-EGAN), a novel Deep Learning (DL) technique, for efficient detection of different types of defects, such as soldering issues, component misplacement and circuit breakages. The data comprises PCB images that represent diverse defect categories. In the preprocessing phase, normalization is applied to standardize the pixel intensity values. Median Filtering (MF) is applied to reduce noise and enhance image quality. A Histogram of Oriented Gradients (HOG) is applied for feature extraction to capture the necessary edge and gradient information crucial for correct defect identification. The ISS-EGAN method uses adaptive random search to optimize the detection process through efficient training and robust defect identification. The performance of the technique is evaluated by several criteria, such as recall, F1-score, and precision. Results showed effectiveness in the detection of high accuracy with low false positives for finding defects and hence offered the fastest and most reliable means of PCB inspection. In modern manufacturing environments, it provides a scalable and effective solution for immediate PCB defect detection by significantly reducing inspection time while ensuring high detection quality.