Detection of Screen Printing Result Using U-Net Convolutional Neural Network Method to Improve Quality Control

Ivan Nur Rahman
Edy Setiawan
Adianto

International Conference on Maritime Technology and Its Application
Vol. 3 No. 1 (2025): ICOMTA : International Conference on Maritime Technology and Its Application

The increasingly competitive convection industry has compelled all enterprises to enhance production quality. One sector affected is the screen printing industry. According to information from one of the companies, customer complaints often arise from suboptimal screen printing results (defects). To achieve satisfactory outcomes, it is crucial to address this issue. Currently, the classification of result eligibility or quality control in screen printing relies on human observation. Pattern recognition technology is significantly transforming the convection industry, particularly within screen printing. This technology enhances quality control, shifting from manual processes to automated quality detection of results. Real-time pattern recognition employs image processing techniques. In this implementation, we utilize image processing with Convolutional Neural Networks for object classification., successfully identifying screen printing defects with an accuracy rate of 97%.