A review of deep learning-based approaches for defect detection in smart manufacturing

Automatic detection of surface faults or defects from images plays a crucial role in ensuring quality control in smart manufacturing. Traditional image processing techniques have limitations in handling background noise, texturing, and lighting variations. To overcome these limitations, the researchers explored deep learning for automated defect identification. The study investigates contemporary mainstream approaches and deep learning methods for flaw detection, highlighting their features, benefits, and drawbacks. The goal is to understand the potential of advanced techniques in enhancing defect identification processes. The research also evaluates the performance of the proposed method and discusses the achievements and limitations of existing defect detection methods. By identifying current challenges, the study aims to pave the way for future advancements in defect detection. It provides an outline to aid the defect detection research community in shaping a new and promising research agenda. Therefore, the study not only presents the proposed method's performance but also offers valuable insights into the strengths and weaknesses of traditional and deep learning-based defect identification approaches.