Vessels Detection in Complex Coastal Area using Hybrid CNN Inspired Dense Connected Model

In this work, we propose a practical and efficient approach for ship detection in remote sensing images.Themethod combines body and ship wake detection by combining deep learning and feature-based visual processing.A deep convolutional neural network (CNN) is used to accurately identify ship bodies, while a feature-driven processing method is developed to detect ship wakes.To enhance analysis, we model the sea area and assess image quality.Duetothe backdrop features like clouds and ocean foam, make it hard to find the actual target ship, it is usually quite difficult to recognize ships in a crowded setting.The suggested process is used to first separate the backdrop from the aim, whereas previous methods were limited to locations with clear conditions.Wake detection is crucial for determining sailing orientation and aiding ship recognition.Notably, it makes it possible to identify ships that are outside the picture limits or obscured by clouds with some awareness.To solve the challenges of identifying ships in crowded backgrounds, a CNN-based model trained on large-scale remote sensing datasets is used.The system is optimized using sophisticated data augmentation approaches to improve generalization over a range of marine environments.A multi-scale feature extraction technique is also incorporated to enhance the detection of ships with varying sizes and orientations. Theproposed method is effective, as demonstrated by experiments on real remote sensing datasets, with identification rates of over 70% for targets.