A cura di MIUR - Direzione Generale per il coordinamento la promozione e la valorizzazione della ricerca

Computer vision and artificial intelligence at the service of marine sciences

Computer vision and artificial intelligence at the service of marine sciences

Computer vision and artificial intelligence will offer innovative tools for use in environmental research in the field of marine sciences. 

An international research group coordinated by the Institute of Marine Sciences of the National Research Council (ISMAR-CNR) has developed an automated method to assess fish abundance in the oceans and its variations over time. 

The study, published in the journal Scientific Reports, involved scientists at Polytechnic University of Marche, Institute for Environmental Protection and Research (ISPRA), Polytechnic University of Catalonia, and Spanish National Research Council (CSIC). 

“We have developed a technique that is based on a supervised machine learning methodology, that is a set of mathematical processes that enable computers to learn to automatically recognize and count individuals photographed in their natural environment or near artificial observatories”, explained Simone Marini from ISMAR-CNR, who coordinated the study. 

The effectiveness of the methodology adopted in recognizing the individuals photographed was validated on 22,000 images of about 176,000 fish. The images were taken at 30 min frequency, by day and night, for a period of two years by the Obsea underwater observatory located off the coast of Barcelona and managed by the Polytechnic University of Catalonia and CSIC. 

The application of these algorithms to thousands of images shows that the method developed by the researchers can be used to accurately track time variations in fish abundance under different operating conditions.

The innovative technique can open new perspectives for the exploration of marine ecosystems. “We will be able to analyse large amounts of underwater images continuously, automatically and in an intelligent way, allowing new approaches also to the conservation of biodiversity”, added the ISMAR-CNR researcher. 

Specifically, the new methodology can be applied to a wide variety of environments: marine protected areas, coastal areas, open sea areas, deeper parts of the oceans. “Technology is also particularly important for monitoring human impacts and the short-term effects of climate change”, Simone Marini concluded. 

Source CNR
Publication date 11/16/2018
Tag Physical Sciences and Engineering