Researchers at the University of Barcelona have developed an artificial intelligence algorithm that detects and assesses the level of plastic pollution in water using air imaging. Now all you have to do is get the drones to fly and autonomously scan the seas and coastal areas, assessing the damage. While it seems impossible to clean everything up, the researchers say tracing the main sources of pollution before it is fragmented and undetectable could help reduce problems right at the beginning of the garbage chain.
To this cybernet digital the team turned to deep learning and began training artificial intelligence on over 3,800 aerial photos of the sea off the coast of Catalonia, using neural networks to improve efficiency. Ultimately, a reliable algorithm was produced to detect and estimate the amount of plastic drifting on the water surface.
- A huge number of photos of the water surface was obtained using drones and airplanes during marine debris monitoring campaigns, as well as experimental studies on known drifting objects, allowed us to develop and test a new algorithm that achieves 80% efficiency in remote detection of floating macro "rubbish," explains one of the authors, Odei Garcia-Garin.
The tool can analyze the photos individually or sort them into different segments by counting the garbage in each section to offer a compaction estimate. In its current form, the tool is a web application available to professionals in the field, but scientists plan to continue working on a version for drones that will automate the entire process. Automatic aerial photography techniques combined with analytical algorithms are more efficient in the control and study of this type of contamination, they add.