Deep Learning-Based Coral Reef Health Assessment Using Modified Inception V3

Marine biologists have accumulated extensive data on underwater environments, yet the analysis of this data is often challenging and labor-intensive. Deep learning has emerged as a powerful tool for interpreting complex data, including imagery. This study introduces a modified InceptionV3deep learning model for classifying coral health conditions (healthy and bleached) using a dataset of 1,169 images. Transfer learning was applied to adapt the pretrained InceptionV3architecture, and hyperparameter tuning was employed to enhance performance. The model achieved an impressive accuracy of 99.14%, significantly improving upon the previous study's highest accuracy of 86%, marking a 13.14% increase. These findings highlight the potential of advanced deep learning techniques, particularly the InceptionV3 model, in accurately classifying coral health conditions. Future research is necessary to validate these results on larger, more diverse datasets and in real-world applications.