Deep Learning Meets the Deep Sea: Automated Coral Classification for Reef Conservation
Coral reefs, essential ecosystems that support diverse marine species, protect coastlines, and boost the fishing and tourism industries, face growing threats from climate change, overfishing, and pollution. Monitoring coral health is crucial for conservation, but traditional methods of classifying coral lifeforms—key to assessing reef health—are often labor-intensive, costly, and invasive.
In their innovative study, Automated Coral Lifeform Classification Using YOLOv5: A Deep Learning Approach, researchers Jannie Fleur V. Oraño, Jerome Jack O. Napala, and Janrey C. Elecito of Southern Leyte State University, in collaboration with Jonah Flor O. Maaghop from Visayas State University, propose a cost-effective solution to this challenge. Published in September 2023, their research leverages YOLOv5, a deep learning model, to automate coral lifeform classification, reducing the need for invasive and time-consuming manual methods.
The team trained the model on a dataset featuring seven coral lifeforms: Branching, Encrusting, Foliose, Massive, Mushroom, Submassive, and Tabulate. Using recall, precision, F1 score, and accuracy metrics, the model achieved a notable 89.29% accuracy rate. This automated approach promises to streamline reef monitoring and aid conservation by providing a non-intrusive, efficient method of assessing coral health.
This article aligns with SDG 14: Life Below Water