Reef Vista: Deep Learning-Powered Underwater Coral Reef Monitoring

Abstract

Coral reefs, vital ecosystems supporting biodiversity and providing economic benefits to millions, face threats that necessitate effective monitoring and conservation efforts. In this study, we present a deep learning-based approach for coral reef monitoring and underwater image enhancement. Leveraging advanced techniques, our model achieves high accuracy in classifying bleached and healthy corals from images captured in diverse underwater conditions. In this paper, we introduce a suite of advanced deep learning-based techniques tailored for coral reef monitoring and underwater image enhancement. These techniques not only augment image quality but also surpass earlier methodologies in terms of effectiveness and accuracy. Key innovations include leveraging Bright Channel Prior for image enhancement and employing state-of-the-art deep learning algorithms for superior results. Our model architecture comprises convolutional layers followed by dense layers with L2 regularization, offering robust performance in distinguishing between coral conditions. We employ data augmentation techniques to enhance model generalization and mitigate overfitting, contributing to reliable predictions on unseen data. Evaluation of the model with various optimizers demonstrates consistent performance across different configurations. Our findings highlight the efficacy of deep learning in coral reef monitoring and underline the importance of leveraging technological advancements for marine conservation efforts.