Machine Learning Approach to CMYK Label Identification Through Screen Angles

Abstract

This research proposes the application of AI to verify the origin of CMYK label that is used in rented machines, focusing on two printing methods: OFFSET and FLEXO. The system employs color patterns and screen color angle verification to ensure sticker authenticity. The paper applies RGB color sensors and imaging cameras, the system converts RGB data into CMYK for label validation. This process incorporates object detection and classification techniques based on deep learning algorithms using Convolutional Neural Networks (CNN) developed through the YOLOV8 framework. The developed model is integrated into an automated system connected to the machinery for real-world applications. Experimental results demonstrate that the system accurately distinguishes between OFFSET and FLEXO printing methods, enhancing compliance with lessor conditions in the labelling industry. This approach further highlights the potential of AI in decision-making and machine control automation.