Classification of vessel types using the visual geometry group based convolutional neural network

Adam Ahmed, Ashraf Adam Ahmad * and Fatai Olatunde Adunola

Department of Electrical and Electronic Engineering, Nigerian Defense Academy, Kaduna, Nigeria.
 
Global Journal of Engineering and Technology Advances, 2024, 21(02), 088–092.
Article DOI: 10.30574/gjeta.2024.21.2.0207
 
Publication history: 
Received on 27 September 2024; revised on 10 November 2024; accepted on 13 November 2024
 
Abstract: 
This study explores the application of Convolutional Neural Networks (CNNs) for the classification of vessel types within Nigerian pilotage districts, focusing on the Visual Geometry Group (VGG) architecture. The research methodology encompasses data collection, preprocessing, model selection, training, and evaluation, resulting in a robust dataset of 15,760 images representing 394 different vessel types. The VGG model achieved a validation accuracy of 94%, alongside a precision of 92%, recall of 90%, and an F1-score of 91%. These metrics indicate strong classification capabilities, yet also reveal potential overfitting, as evidenced by a plateau in validation accuracy after 50 epochs. The confusion matrix highlights the model's challenges in accurately classifying certain vessel types, suggesting a need for further refinement. Overall, this work contributes to the growing body of knowledge in maritime Artificial Intelligence (AI) applications, with implications for improved operational efficiency and safety in port management.
 
Keywords: 
Convolutional Neural Networks (CNN); Visual Geometry Group (VGG); Pilotage Districts; Artificial Intelligence (AI); Port Management
 
Full text article in PDF: