AI Models for Defect Detection in Lean Manufacturing: A Comparative Study of Deep Learning Techniques

In this comparative study, we evaluate deep learning techniques for defect detection within lean manufacturing settings. Our methodical literature review identified key deep learning architectures—CNNs, R-CNNs, and YOLO variants—for their track record in accurate defect identification. Employing a standardized dataset, models were assessed against lean manufacturing criteria: accuracy, precision, recall, and processing speed. The analysis revealed that while DCNNs offer high precision in detecting surface defects, R-CNNs decrease false negatives, vital for production reliability. Models like FCNNs show promise for rapid defect recognition, an essential component for maintaining manufacturing throughput. Our findings suggest these techniques can significantly contribute to the goals of lean manufacturing by enhancing defect detection and optimizing quality control processes. Future research should continue to refine AI model integration and scalability within varied manufacturing environments, emphasizing adaptability and efficiency to support lean manufacturing principles. This study provides a foundational understanding of AI's potential in improving defect detection, a step towards integrating cutting-edge technology in manufacturing systems for enhanced quality assurance and waste reduction.