As millions of tons of plastic wash into the ocean everyday, scientists have their work cut out for them in trying to keep tabs on its whereabouts, but they may soon have a useful new tool at the their disposal. Researchers at the University of Barcelona have developed an algorithm that can detect and quantify marine litter through aerial imagery, something they hope can work with drones to autonomously scan the seas and assess the damage.
Taking stock of our plastic pollution problem is a tall order, with so much of it entering the ocean each day and being broken down into smaller fragments that are difficult to trace. The University of Barcelona team has taken aim at those pieces floating on the surface, hoping to improve on current methods of tracking their distribution, which involve surveying the damage from planes and boats.
An interesting example of this is the work carried out by The Ocean Cleanup Project, which has ventured into the Great Pacific Garbage Patch with research vessels and flown over the top of it with aircraft fitted out with sensors and imaging systems. Most recently, it demonstrated a way of doing this using infrared to distinguish pieces of plastic swirling about in the ocean from other ocean debris.
The University of Barcelona team has instead turned to deep learning techniques to analyze more than 3,800 aerial images of the Mediterranean off the coast of Catalonia. By training the algorithm on these photographs and using neural networks to improve its accuracy over time, the team wound up with an artificial intelligence tool that could reliably detect and quantify plastic floating on the surface.
“The great amount 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 a 80 percent of precision in the remote sensing of floating marine macro-litter,” says team member Odei Garcia-Garin.
The tool can analyze images individually or sort them into different segments, tallying up the litter in each section to offer an estimate of density. As it stands, the tool is an open-access web app available to professionals in the field, but the team expects to develop a version that can work with drones, to fully automate the process.
“Automatic aerial photography techniques combined with analytical algorithms are more efficient protocols for the control and study of this kind of pollutants,” says Garcia-Garin.