Revolutionizing Quality Control: A Machine Learning Approach for Precise Surface Defect Detection and Localization in Industrial Components

This paper focuses on the creation of a machine learning application designed to efficiently identify and categorize surface imperfections in a variety of components. Conventional methods, relying on visual examination or manually crafted characteristics are prone to inaccuracies and are time-intensive. Leveraging the capabilities of Artificial Intelligence architecture, we not only achieve precise defect classification but also accurate defect localization. In contrast to traditional approaches, Artificial Intelligence can adeptly learn optimal visual features in a supervised manner. In real image data set we performed our experiments which illustrate that our approach outperforms current methods while maintaining computational efficiency. We have developed a machine learning model for surface defect detection in steel flat surfaces through image segmentation. This advancement has resulted in reduced defects, heightened manufacturing accuracy, increased production rates and enhanced product quality.