Advanced machine learning based surface defect detection systems

This paper presents a smart approach to surface defect detection utilizing machine learning algorithms and image processing techniques. The integration of discrete wavelet transform (DWT) and the Otsu binary algorithm enhances image quality and effectively segments defects. Through extensive experimentation, we establish a combined bivariate performance measure (CBPM) to evaluate the effectiveness of various machine learning models, focusing on their accuracy and training efficiency. Our findings highlight significant improvements in real-time defect identification, which is critical for maintaining quality control in manufacturing processes. The proposed system not only optimizes performance metrics but also promotes early defect detection, leading to enhanced production quality and reduced operational risks. This work contributes to the growing field of automated defect detection, showcasing the potential of machine learning in improving industrial practices and ensuring product safety.