Artificial intelligence protects the oceans by identifying ships sinking bilge water

While cases of ships unloading and bilge water pouring off the coast are reported regularly, a startup has decided to use satellite imagery to condemn polluters. Their best ally is artificial intelligence, which was developed with the help of Amazon.

This nonprofit organization, called SkyTruth, tracks boats, whether tankers or boaters, who drop their sewage. These can actually disturb marine flora and fauna when they are tossed off the coast. A prohibited method, too, explains the International Maritime Organization, the International Maritime Organization, which is dependent on the United Nations.

Since it is almost impossible to monitor a ship’s activity relying on the collaborative work of volunteers alone, SkyTruth with the help of Amazon Web Services has developed artificial intelligence capable of interpreting satellite data and indicating the date and time of the boat’s discharge of sewage. The algorithm is based on radar data from Copernicus Sentinel-1, the observation satellites that the European Space Agency (ESA) has put into orbit. Imaging makes it possible to see the long traces of bilge water on the surface with the naked eye. Using a deep learning method, SkyTruth then trained an AI to identify the source of this marine pollution.

Artificial intelligence to aid research

Ship owners and owners can be denounced and held responsible for this pollution. The idea is basically to put the sword of Damocles over the heads of the polluters by indicating what ships leave in their wake

This is not the first time Amazon Web Services has used IT infrastructures to combat human impact on the environment or to advance science. With Capgemini and researchers, AWS has developed artificial intelligence capable of recognizing sperm whales in order to better understand their migration flow and encourage ships to reroute their sea paths. Along with the World Wide Fund for Nature, Amazon has also created an orangutan identification system to identify its population.