Marine Vision-Based Situational Automatic Ship Detection Using Remote Sensing Images

Object detection or identification is the one of the fundamental problems in computer vision application. Even though, object detection is a successful research area, detection of small object from remote sensing images is complicated. Remote sensing image-based automatic ship detection is part of marine surveillance system. Marine safety is also one of the main security sectors for national security. So, to avoid the risk of pirates and extremists entering the harbor zones, early detection of ship is necessary. Similarly, when there are accidents of ships in maritime, identifying the ship is a challengeable task. So, when considering oceanic security and safety, automatic detection of ship is obligatory. The deep learning model, particularly MobileNet, without forgetting architecture, was considered for automatic and early ship detection. The experimental setup produced the 98.2% accuracy rate with Kaggle ship dataset. The experimental setup is evaluated with the performance analysis and finally compared with some other techniques.