Masterarbeit
Synthetic Aperture Radar-Based Ship Detection Using Dual Object Detection
Completion
2025/07
Research Area
Students
Nithinkumar Mirle Prasannakumar
Advisers
Abubaker Gaber
Prof. Dr.-Ing. Martin Gaedke
Description
The marine sector frequently faces numerous illicit activities, such as unauthorized fishing near coral reefs, which results in the devastation of marine biodiversity. Other challenges include unlawful immigration, oil leaks from vessels, and illicit shipping. To counter these activities, the development of satellite remote sensing technology, especially Synthetic Aperture Radar (SAR) sensors, has significantly enhanced surveillance capabilities. Freely available SAR data still poses a challenge for the detection of smaller vessels accurately. Furthermore, tracking vessel paths via satellite imagery is still a complex task. A potential remedy could be the ongoing expansion of AI models through training on newly available data. This method could offer substantial advantages by enhancing the precision of vessel detection and path monitoring.
In addition, the creation of lightweight AI models designed for real-time or near-real-time applications could also be beneficial. These streamlined AI models could boost the effectiveness of monitoring and detection procedures, especially in scenarios where immediate insights are vital. The objective of this thesis is to find an approach to solve the above problem in the context of ship detection based on SAR images, trained using object detection algorithm. This includes the state of the art analysis regarding these fields. The demonstration of feasibility with an implementation prototype of the concept is part of this thesis as well as a suitable evaluation on a representative test dataset including quality measurements.
