Cienciaes.com: Artificial intelligence to detect floating garbage in the sea. We spoke with Odei García Garín.

2021-02-16 10:51:48

We generate garbage at such a rate that there is practically no place on Earth that is not invaded by it in one way or another. A very high percentage of all this garbage is made up of low-density plastic materials that, on many occasions, are washed away by water and dumped into seas and oceans. Once there, a part of the waste is deposited on the seabed and another is dragged by the ocean currents that surround the planet. In the great oceans, the currents follow circular paths that confine the floating materials in very specific regions, creating immense islands of floating materials. In the Pacific, to give a dramatic example, between the coasts of Hawaii and California, there is the largest of these islands, occupying an area of ​​1.6 million square kilometers, that is, more than three times the total area of ​​Spain.

Such agglomerations of debris are causing problems for marine ecosystems around the world and, by extension, for us as well. There is plastic waste of all sizes. as he tells us Odei Garcia Garinour guest on Talking to Scientists, scientists subdivide this marine debris into “nanoplastics” (those with dimensions of several nanometers), microplastics (between 1 micron and 5mm), mesoplastics (from 5mm to 2.5cm) and macrotrash. (>2.5cm).
Such an abundance of waste is putting the scientific community on alert and there are many researchers who strive to develop methods that allow its study, with the hope that one day their research work will serve to develop strategies to fight against all this contamination. floating or submerged.

Odei García Garín is currently developing his doctoral thesis and has just published an article in the Environmental Pollution magazine that talks about the application of artificial intelligence to monitor floating plastics in seas and oceans. The project is called MARLIT and it is an open access web application based on an algorithm designed with deep learning techniques, that is, techniques that, through autonomous learning, allow the detection and quantification of floating plastics in the sea.

Logically, the identification of the garbage that floats or sinks in seas and oceans is not easy. To the great diversity of materials that exist in very different shapes, colors and compositions, we must add the movement of the waters that transport them, where the waves, the wind or the reflections of sunlight interfere with their location.

Odei and his colleagues from the Faculty of Biology and the Biodiversity Research Institute of the University of Barcelona have analyzed more than 3,800 images obtained through direct observations from ships, planes or drones, in areas of the Mediterranean coast of Catalonia and Valencia to “train ” to the algorithm. Thus, he has learned to assess the existence, abundance or distribution of plastic pollutants in the sea with a reliability of 80%.

in the development of MARLIT researchers from the Large Marine Vertebrates Group of the University of Barcelona (ub) and IRBio, the Biostatistics and Bioinformatics research group (GRBIO) and the integrated group of the Barcelona Bioinformatics platform.

I invite you to listen to Odei García Garín, predoctoral researcher in the Department of Evolutionary Biology, Ecology and Environmental Sciences of the Biodiversity Research Institute from University of Barcelona.

References:
Odei Garcia-Garin et al., Automatic detection and quantification of floating marine macro-litter in aerial images: Introducing a novel deep learning approach connected to a web application in R Environmental Pollution Volume 273, 15 March 2021, 116490

MARLITan application based on artificial intelligence to study floating marine macro-debris