Automated Algae Detection and Quantification on Coral Reefs Using U-Net Segmentation Model

Coral reefs, essential marine ecosystems, face increasing threats from algae overgrowth, leading to coral bleaching and ecological imbalances. Efficient monitoring and quantification of algae presence are crucial for coral conservation. This study introduces a segmentation model leveraging the U-Net architecture to detect and quantify algae coverage on coral reefs. The model was trained on a custom dataset created using underwater videos provided by the International Union for Conservation of Nature (IUCN). Extracted samples underwent pre-processing and data augmentation to enhance the dataset’s diversity and size. The proposed segmentation model achieved satisfactory accuracy, with a mean Intersection over Union (IoU) of $\mathbf{0. 8 5}$, indicating reliable accuracy in distinguishing algae and coral. This level of accuracy enables practical use in conservation efforts, offering a competitive advantage over traditional manual monitoring methods. Segmentation results were further processed to calculate the percentage coverage of algae overlapped on coral using an image processing technique. The model’s output is designed for integration into a mobile application, providing field researchers and marine enthusiasts with real-time analysis and quantification of algae coverage. The research signifies a substantial advancement in the application of deep learning techniques for environmental monitoring, contributing significantly to marine biodiversity preservation efforts.