Machine Learning and Deep Learning Based Coral Species Classification: A Survey

Abstract

Marine ecosystems cannot function without coral reefs. There have been reports of coral reefs declining in health and population recently. The amount of information on coral reefs available for study has expanded dramatically because to underwater imaging tools like autonomous underwater vehicles (AUV) and towed diver sleds. Variations within classes are large, class boundaries are complex, and picture clarity varies greatly in coral reefs. Because of this, categorizing corals is difficult. In order to better understand coral reefs, this study provides a comprehensive overview of how machine learning and deep learning have been applied to underwater picture processing. Image categorization, object detection, and scene understanding are just a few of the many computer vision problems where deep learning approaches have produced state-of-the-art results. From a computer vision standpoint, marine ecosystems are challenging because of the scene complexity involved. Deep learning has the ability to effectively solve the difficult task of classifying coral species.