Marine Vessel Image Recognition Based on Optimized Meta-Heuristics Algorithm with Deep CNN Architectures

Abstract

Marine vessels are vital for international transportation and defense which require effective classification methods for tracking and security. Even though already existing traditional classification algorithms and models are faced with environmental changes, occlusions, and being able to extract features is still limited. To get around these difficulties, the study integrates the Ant Colony Technique (ACO) with deep Learning convolutional networks (CNNs), particularly ResNet-50 and VGG-16, for the higher quality of marine ship classification. The process consists of the preprocessing of a Kaggle dataset containing 9,310 cargo, military, cruise, carrier, and tanker ship images. The image is prepared by resizing, normalizing, and augmenting and then by using deep learning models to classify them. These discoveries are the best proof of the success of combining deep learning with meta-heuristic optimization under marine vessel classification, thus ensuring the highest precision and efficiency in maritime applications.