Monitoring Coral Reefs Death Causes with Artificial Intelligence

Coral reefs play key roles in the marine ecosystem, providing nursery, refuge and nurturing areas for many organisms. However, they are in danger from invasive species like Crown-of-Thorns and massive coral bleaching caused by rising water temperatures. In this paper, an innovative approach for reef monitoring is proposed based on Machine Learning methods like Naive Bayes, Decision Tree, KNN, SVM, Random Forest and XGBoost to automatically classify corals into varying bleaching severities by training on past bleaching events. The experiment was reinforced using SMOTE and optimisation algorithms such as Grid Search, PSO and GA. It was found that XGBoost produced a higher accuracy after balancing the training dataset (80.11%) and Random Forest performs better with PSO(77.8%). Furthermore, Deep Learning was used to detect Crown-of-Thorns in underwater images using a custom trained EfficientDet-D0 which yielded 81% of correct detection. These novel methods are aimed at assisting marine scientists in protecting reef ecosystems.