This research tackles maritime safety challenges by using AI-generated multi-weather imagery to improve object classification systems without human supervision.

• Created AIMO dataset with balanced object categories across diverse weather conditions using generative AI
• Developed a novel domain adaptation approach that improves maritime object recognition across varying weather conditions
• Achieved significant performance gains by addressing long-tail data distribution problems in both object categories and weather conditions
• Enhances maritime security capabilities by improving detection reliability in challenging environments

This advancement enables more robust maritime monitoring systems that maintain accuracy across changing weather conditions—crucial for maritime safety, surveillance, and environmental protection.

Domain Adaptation from Generated Multi-Weather Images for Unsupervised Maritime Object Classification