Underwater Detection of Ship Hull Biofouling Using Computer Vision

Abstract

This study proposes a simulation-based approach for biofouling detection on ship hulls using YOLOv8-Nano. The integration of deep learning-based object detection for real-time biofouling detection demonstrates potential in reducing maintenance costs and improving ship performance. YOLOv8-Nano effectively detects biofouling organisms such as barnacles, mussels, and algae in underwater environments, even with challenges like varying visibility and object sizes. The research highlights the feasibility of using automated detection for biofouling management, offering a scalable solution compared to traditional methods like dry-docking and manual cleaning. However, the study is based on a simulated environment, and real-world testing is required to validate the system’s operational effectiveness. While the model performs well for larger organisms, challenges remain in detecting smaller or partially obscured biofouling due to environmental factors such as lighting and water clarity. The findings suggest future improvements, including enhancing model accuracy with multispectral imaging, refining the detection capabilities, and integrating AI-driven predictive analytics for proactive biofouling management. This work lays the foundation for the development of an efficient and scalable biofouling management system, contributing to sustainable maritime maintenance practices.