P‐50: AI‐Empowered Display Industry: Innovative Breakthrough in Defect Inspection

Based on artificial intelligence technology and using deep learning image classification principles, factory product defects are automatically classified. The AI model algorithm is upgraded and empowered to support intelligent image retrieval, batch automatic annotation, and small sample generation technology to assist defect detection, achieving a closed‐loop management and operation of defect data and hardware. This paper innovatively adopts three innovative measures: self‐developed algorithms based on U‐Net structure, self‐developed display large models based on VIT architecture, and self‐developed high availability K8S monitoring and scheduling SDK. In addition, the use of machine learning intelligent factor analysis algorithms enables AI big data agile analysis to identify commonalities, display analysis results with one click, and improve the effectiveness of AI intelligent monitoring based on attribution classification. Therefore, our independently developed ADM system is the first to achieve efficient process closed‐loop for intelligent defect detection, analysis, stop loss, root cause identification, and improved monitoring, promoting the positive management of process quality management in the LCD industry and achieving a spiral upward trend.