Coral reefs, covering only 0.1% of the ocean, support over 25% of marine productivity and are crucial for biodiversity, coastal protection, and human resources. However, climate change, ocean acidification, and pollution threaten these ecosystems, impacting coral growth and leading to bleaching. Monitoring coral health is essential for sustainable ecosystem and fishery management. Traditional survey methods are time-consuming and inefficient for large-scale assessments. This project proposes an automated coral recognition system using machine learning. The system will analyze underwater images to accurately identify coral species and their distribution, enabling efficient quantification of coral abundance and coverage. The system will enhance coral coverage calculations and facilitate rapid assessments across various environments. Coral coverage, the proportion of area covered by coral, is a key indicator of reef health. Traditional methods require significant manpower and time, limiting their scope. The proposed system will use unmanned underwater vehicles to capture large-scale imagery, with machine learning and image processing techniques to quickly identify and calculate coral coverage. After establishing the recognition system, 3D models of coral reefs will be created to improve accuracy in coverage calculations. The study will use Structure from Motion (SfM) to build 3D models and analyze coral reef structure differences under various human disturbances. This will help quantify shelter volumes and assess the relationship between coral coverage and reef function. This research will provide an efficient and accurate method for tracking coral growth, supporting coral protection efforts, and aiding ecological monitoring in Taiwan. It aims to enhance the scientific management of coral reef protected areas and contribute to the sustainable management of coral ecosystems.