Researchers have trained an AI algorithm to detect plastic marine litter.

Plastic and other refuse float on the surface of many of our precious seas and oceans, endangering the conservation of marine ecosystems. Conventional anti-pollution measures rely on observations of this floating junk from boats or planes. However, due to physical constraints, these efforts are inadequate in the face of mounting marine pollution. But hope is at hand from MARLIT, an app based on an artificial intelligence-based algorithm developed by researchers at the University of Barcelona (UB). (EurekAlert)

MARLIT: How it works

Experts of the Faculty of Biology and the Biodiversity Research Institute of the University of Barcelona (IRBio) have developed MARLIT, an open-access web app that functions based on an algorithm trained on deep learning.

They designed a new algorithm and trained it on more than 3,800 aerial images of the Mediterranean coast in Catalonia. Their objective was to automate the quantification of floating plastics in the sea using deep learning techniques and artificial neuronal networks.

“The great number of images of the marine surface obtained by drones and planes in monitoring campaigns on marine litter -also in experimental studies with known floating objects- enabled us to develop and test a new algorithm that reaches an 80% of precision in the remote sensing of floating marine macro-litter”, said Garcia-Garin, member of the Department of Evolutionary Biology, Ecology and Environmental Sciences of the UB and IRBio.

MARLIT enables analysis of the sea-borne plastic litter images individually, as well as segment-wise if so desired by the user. It also gives an estimate of the density of the garbage by referring to the image metadata (namely height, resolution).

According to the researchers, it could be possible in the future to use the app with remote sensors mounted on a drone. This could make monitoring sea waste more expansive, quicker, and reliable.

Related Story:  The Norwegian AI Lab Helps Salmon Aquaculture