A review of coral reef classification study using deep learning approach

Abstract

Corals are critical as indicators of the health and diversity of the marine ecosystem. Corals, on the other hand, are complex organisms to study due to their sensitivity to changes in temperature, acidity, and pollution. Marine ecosystems frequently have complex scenes, making them difficult to monitor manually. Automated technology is required to monitor the health of our oceans and to aid in the efficient detection and identification of coral reefs. Manually covering the entire research region takes a long time, especially when it comes to the coral classification procedure. It was a typical sequential activity that took longer to complete, was more expensive, required human participation, and produced results with a low degree of accuracy. The researchers are looking for a new method for automatically classifying significant image data of coral reefs with a faster computation time and a higher degree of accuracy, with little or no human intervention. This review paper emphasized the use of deep learning as a method for classifying coral reef components. The deep-learning system may gradually acquire the ability to extract high-level features, obviating the need for expert knowledge. As a consequence, deep learning outperformed conventional methodology in studies of coral reef classification. The study also emphasized the manual methodology, particularly for the classification of coral reefs. Implementing a deep learning methodology has been shown to reduce costs, increase accuracy, and monitor changes in marine ecosystems. This paper will present the results of a previous study on coral reef classifications using various types of deep learning approaches. The purpose of this review paper is to share and combine successful findings from previous research in order to demonstrate the deep learning approach's superior performance in coral reef classification research.