Multi-Model System for Sustainable Coral Reef Conservation in Sri Lanka

Coral reefs, often called the “rainforests of the sea”, are vital ecosystems supporting 25% of all marine species and providing significant ecological, economic, and coastal protection benefits. However, they are under severe threat from climate change, pollution, and overfishing, leading to widespread degradation. This research presents an automated framework designed to enhance the efficiency of ongoing coral conservation efforts. The proposed methodology integrates advanced technologies, including Convolutional Neural Networks (CNNs) with cross-domain transfer learning for accurate species identification and growth monitoring. A hybrid scheduling algorithm, incorporating AHP, TOPSIS, and Reinforcement Learning (RL), optimizes task prioritization. Resource and labor allocation is refined using Explainable AI (XAI), while a Knowledge Graph Convolutional Network (KGCN) enhances real-time recommendations. This approach achieves 94.2% accuracy in species identification, 85% accuracy in growth monitoring, optimizes task prioritization with a 95% task completion rate and 90% resource utilization efficiency, while real-time recommendations achieve an AUC of 0.964 and an F1 score of 0.921. These results significantly improve the precision, adaptability, and efficiency of coral restoration activities, ensuring more sustainable and responsive conservation strategies.