Ki researchers develop software to detect plastic debris in the ocean

Ki researchers develop software to detect plastic debris in the ocean

Barcelona, Feb. 10, 2021

Researchers at the University of Barcelona have developed AI software that uses aerial imagery to detect plastic debris in the sea. The deep-learning models were trained on more than 3,700 images from drone and aircraft imagery of the Mediterranean Sea off the coast of Catalonia.

In their study, researchers Morgana Vighi, Odei García-Garin and Bertrand Bouchard describe how their algorithms can both detect and quantify marine debris in images. Their freely available web app allows users to identify this trash in images of the ocean surface, the researchers said.

Their proposed algorithm is based on a deep-learning approach that uses convolutional neural networks, which can learn from unstructured (or unlabeled) data.

In tests, the algorithm was able to remotely detect 80% of what the team calls „floating marine debris.“ The team is now working to develop an app that can work with aerial drones to fully automate the counting process.

An estimated 10 million tons of plastic enter the world’s oceans each year, causing an „unprecedented environmental crisis,“ according to a 2019 study.