This project is a collaboration between the Marine Angiosperm Ecology Group of the Spanish Institute of Oceanography (IEO-CSIC) and the Machine Learning area of the University of Oviedo. Its objective is to develop a deep neural network-based system for the automatic segmentation of Nanozostera noltei meadows in drone-captured images of the Villaviciosa estuary.
Nanozostera noltei meadows are essential marine ecosystems that provide important ecological services, such as coastal protection and carbon sequestration. Due to their ecological value, they are protected by national and international regulations, and their monitoring is key to assessing the impact of conservation measures.
Currently, the segmentation of these meadows is performed manually, a process that is slow, costly, and not scalable. Operator dependence introduces variability in the results, and manual segmentation is not feasible for large volumes of images.
The use of deep neural networks is proposed to automate segmentation, improving the accuracy, speed, and scalability of the process. This approach will efficiently identify Nanozostera noltei meadows, overcoming the limitations of the current manual method.