Impact of Sea Fog on Warship Classification Using YOLOv8

Sea fog is a critical weather condition that significantly affects the operational capabilities of warships. Due to the diverse missions, sensors, and weapons systems across different types of warships, accurate type classification is essential. This study investigates the effect of sea fog on the classification accuracy of various warship types. A dataset containing 944 images of 10 distinct South Korean warship classes was constructed, and the YOLOv8 deep learning model was trained using advanced image preprocessing and data augmentation techniques. To simulate foggy conditions, image filtering and mix-up techniques are employed, allowing for a comparative analysis of model performance under varying degrees of simulated fog. Experimental results indicate that while the model performed well in clear conditions, its accuracy progressively declines with increasing fog density, with a significant drop in performance observed at 40% fog simulation. This study underscores the significant impact of sea fog on warship type classification and establishes benchmarks for evaluating the feasibility of autonomous warship operations under varying fog conditions. Furthermore, the methodologies developed in this research can be applied to the classification of warships from other nations, extending its applicability beyond the scope of this study.