Classification of Coral Reef Species using Computer Vision and Deep Learning Techniques

Coral reefs are among the most diverse and productive ecosystems, teeming with life and providing many benefits to marine life and human communities. Coral reef classification is popular for many important reasons, such as assessing biodiversity, prioritizing conservation actions to protect vulnerable species and their habitats, and many other objectives related to scientific research and interdisciplinary studies on marine ecosystems. Classifying images of coral reefs is challenging due to their great diversity and subtle differences in morphology. Manually classifying them is a time-consuming process, especially when dealing with large datasets. This can limit the scalability and efficiency of scientific research and conservation efforts. This study proposes an automated classification approach using computer vision and deep learning techniques to address these challenges, employing models such as YOLOv5l, YOLOv8l, and VGG16 to classify images of coral reefs. The dataset, comprising 1,187 images of five coral species, was augmented for robustness. YOLOv8l demonstrated superior performance with an accuracy of 97.8%, significantly outperforming the other models in terms of speed and accuracy. These results demonstrate the potential of advanced deep-learning models to improve coral reef monitoring and conservation efforts. This approach aims to streamline classification processes, improving the efficiency and scalability of coral reef research and conservation initiatives worldwide.