Our website is currently undergoing maintenance. Article records and full texts remain available. Submissions can be made through the Editorial Manager.
Deep Learning for Ceramic Surface Defect Inspection: A State-of-the-Art Survey
B.S.Wang, L. Zhang
Abstract
Surface defect detection in ceramic manufacturing is crucial for ensuring product quality and functional performance. This comprehensive review examines the evolution and current state of deep learning applications in ceramic surface inspection. Traditional inspection methods, including manual visual inspection and conventional image processing techniques, face limitations in handling complex textures, varying lighting conditions, and subtle defect patterns. The review analyzes various deep learning architectures, from basic Convolutional Neural Networks (CNNs) to sophisticated region-based networks and segmentation models, highlighting their strengths and limitations in ceramic defect detection. Implementation considerations, including data acquisition, preprocessing strategies, model training approaches, and real-time processing requirements, are discussed extensively. The paper addresses critical challenges such as limited labeled data availability, class imbalance, and the need for robust performance under varying production conditions. Recent innovations in network architectures, training methodologies, and deployment strategies that address industry-specific challenges are examined. The review provides valuable insights for both researchers and practitioners, bridging the gap between academic research and industrial implementation while identifying emerging trends and future research directions in this rapidly evolving field.
Keywords
Automated inspection, surface quality control, machine vision, neural networks, manufacturing automation
Göller Verlag GmbH · info@goeller-verlag.de · Site Notice · Privacy · Terms