AI Infused Underwater Analysis for Enhanced Coral Species

Among the most widely and varied ecosystems on our planet, coral reefs are yet more susceptible to the changes spurred by climate change, pollution, and human activities. Effective monitoring and a deep understanding of health changes in time are necessary elements for conserving these ecosystems. For the first time, we are presenting here a novel system integrating advanced deep learning techniques with computer vision to automatically detect, segment, and track coral reefs using images taken underwater. In this paper, we trained several models, namely YOLO, U-Net, DeepLabv3+, and FCN, on different sets of the selected coral reef images, demonstrating their feasibility for monitoring the health indicators of corals correctly. This approach is unique in combining the models optimized for different tasks; that is, while the model of YOLO v10 excels in detecting the coral object, the U-Net performs outstandingly in fine details and fullscale evaluation of coral structures. Furthermore, real-time monitoring integration would help scale and make coral conservation more efficient, thus enabling better decision-making in protecting these incredible marine habitats.