Deep Learning-Based Ship Detection: Enhancing Maritime Surveillance with Convolutional Neural Networks

It has great importance in maritime surveillance in preventing illegal activities and safeguarding oceanic operations. However, the conventional processes for ship detection fall under an inefficient category because they depend highly on manual monitoring. This study explores deep learning, particularly Convolutional Neural Networks, for the automation of detection using satellite images. A custom CNN model was designed and trained on a publicly available dataset to classify images as "ship" or "no-ship." The architecture consists of convolutional and max-pooling layers for feature extraction followed by dense layers for classification with dropout techniques to prevent overfitting. Data augmentation techniques were used to improve the generalization of the model. The model achieved an accuracy of about 86%, which is promising for real-world maritime surveillance applications. Further improvements, including hyperparameter tuning and better balancing of the dataset, could give even better performance. The results confirm that the CNN-based ship detection method represents a promising way to automate ocean monitoring, thereby improving the accuracy and efficiency of maritime security operations.