Application of Deep Learning in the Classification of Maritime Safety Information

Maritime safety information (MSI) refers to urgent information concerning navigation warnings, weather warnings, weather forecasts, and other safety-related information for navigation. MSI is primarily disseminated through two systems: the international NAVTEX (Navigational Telex) system and the Enhanced Group Calling (EGC) system. NAVTEX receivers on ships can automatically receive MSI, enhancing safety of life and property at sea and reducing the workload for communication personnel. Due to the narrowband direct printing telegraphy (NBDP) technology used in the broadcast of MSI, the information must be concise, it makes the information difficult for navigators to interpret. To address this challenge, various machine learning solutions have been proposed. Among these, deep neural networks have shown superior performance in MSI classification. This study provides a more detailed analysis of deep neural networks applied to MSI classification. It utilizes collected MSI datasets, including thousands of navigation warnings, and compares the performances of different models based on accuracy, precision, recall, and F1-score.