Enhancing the Automatic Recognition Accuracy of Imprinted Ship Characters by Using Machine Learning

In this paper, we address the challenge of ensuring safe operations and rescue efforts in emergency situations, for sake a sustainable marine environment. Our focus is on character recognition, specifically deciphering characters present surface aged corroded ships, where markings may have faded or become unclear over time, contrast to vessels with clearly visible letters. Imprinted ship encompassing engraved, embroidered, other variants found components serve as vital markers identification, maintenance, safety technology. The accurate recognition these essential efficient effective decision making. This study presents machine-learning-based method that markedly improves accuracy imprinted numbers characters. improvement achieved by enhancing data classification through augmentation. effectiveness proposed was validated comparing it State-of-the-Art technologies within dataset. We started originally sourced dataset then systematically increased size, using most suitable generative adversarial networks our compared classic convolutional neural network (CNN)-based classifiers classifier, CNN-based classifier (CNN-ISC). Notably, augmented dataset, CNN-ISC model impressive maximum 99.85% 99.7% alphabet digit respectively. Overall, augmentation improved digits alphabets, outperforming methods.