Real-time Quality Monitoring System for the Apparel Industry in Sri Lanka

This paper introduces SeamSense, an innovative real-time quality monitoring system aimed at transforming the apparel industry's quality control process in Sri Lanka. By automating the traditionally manual quality inspection, SeamSense targets the detection of seam defects using high-definition video capture, fog computing, and advanced machine learning models, particularly You Only Look Once 8 and You Only Look Once 9. These models are optimized for real-time seam defect identification, focusing on high-low and open-seam defects through advanced feature extraction and rapid processing. SeamSense is projected to reduce seam defects by approximately 30-40%, thus enhancing overall production quality. Beyond defect detection, SeamSense achieves optimized database management, strengthened security, and improved defect prediction accuracy. This is achieved through techniques such as motion-adaptive frame extraction, model-based augmentation, and fusion models. The fusion model uniquely integrates traditional machine learning with time-series analysis, enabling a comprehensive approach to quality control that supports consistent and high-standard apparel production. In doing so, SeamSense drives significant improvements in efficiency and innovation across the garment manufacturing sector.