AI-Enabled Computer Vision for Zero-Defect Manufacturing: A Scalable Architecture for Industrial Resilience and National Competitiveness

Abstract

The global manufacturing sector is undergoing a paradigm shift toward zero-defect manufacturing (ZDM) to meet the demands of Industry 4.0, which prioritizes precision, efficiency, and sustainability. This paper presents a scalable architecture for AI-enabled computer vision (CV) systems designed to eliminate surface defects in industrial production lines. By integrating modular design principles, federated learning frameworks, and human-in-the-loop (HITL) optimization, the proposed system addresses critical challenges in scalability, data privacy, and economic feasibility. Technical innovations include hybrid CNN-Transformer models achieving 99.2% defect detection accuracy, federated learning protocols reducing data latency by 40%, and a cost-benefit model demonstrating a 22% return on investment (ROI) over five years. Validated against ISO 9001 standards, this architecture enhances supply chain resilience and positions nations competitively in advanced manufacturing.