Coral Detection using Artificial Neural Networks based on Blurry Images for Reef Protection in Cayo Blanco, Honduras

Knowing the implications and benefits of coral reefs, the identification, monitoring and protection of coral species are of major importance, and applying technological advances to this process greatly adds value. Technology allows for better efficiency in terms of time, resources, personnel and the gathering of data. In Santa Fe, Honduras recently the discovered of a coral reef called Cayo Blanco was made, which is a continuation of the Mesoamerican reef. A neural network was trained using approximately 30% of blurry images. This research aims to create a graphic user interface equipped with a neural network capable of counting, classifying, and identifying at least five coral species found in Cayo Blanco, Honduras. The algorithm has 95% precision, is trained with 399 images in the Coral database, can show and register the detections, and provide a specific measurement of confidence. We concluded that for machine learning models, the quantity outperformed the quality of the image data.