Virtual Booth

Ship hull inspection is important for maintaining the safety and performance of marine vessels; however, traditional diver-based inspections are risky, costly, and time-consuming. This project develops an Autonomous Underwater Vehicle (AUV) for real-time ship hull inspection using deep learning and computer vision. The system integrates a Raspberry Pi, underwater camera, sensors, and propulsion motors for autonomous navigation and live video capture. A YOLO–ResNet-50 framework with Genetic Algorithm (GA) optimisation is used for defect detection and classification. The optimised ResNet-50 achieved 90.00% accuracy, while the best YOLO configuration achieved 80.39% accuracy, providing a safer and more efficient maritime inspection solution.