AI dives into a sea of data, from plankton to pollution

Summary

By Tom Cassauwers When asked why Jean-Olivier Irisson, a scientist at Sorbonne Université in Paris, decided to dedicate his life to studying microscopic creatures in the sea, his answer was simple: “They are beautiful.” Beauty may not be the first thing that comes to mind when we think of plankton – organisms that drift in water and come in an extraordinary variety of shapes and sizes. But images by Irisson’s team tell a different story. Shown in striking blues and oranges, as well as black and white, they reveal an unfamiliar and strangely beautiful world. “This one served as the model for the head of the creature in the Alien movie franchise,” Irisson said, pointing to one particularly unusual specimen. Plankton are not just fascinating to look at. They are fundamental to marine ecosystems, and play a crucial role in regulating Earth’s climate by helping remove carbon dioxide from the atmosphere. “Half of the photosynthesis in the world is done by plankton,” said Irisson. “They also form the foundation of the food chain at sea. Without them, there would be no fish.” Yet studying plankton is painstaking work. Researchers collect water samples and examine them under microscopes. They can contain huge numbers of organisms that all need to be identified and classified – a slow and demanding task when done manually. AI is now helping to speed things up. “We already use software to capture and process images, to identify every individual organism,” said Irisson. “For a single dataset we have 2.2 million of these images of individual organisms. That’s impossible to classify manually. With AI, we can, however, do this very fast. That’s opening up new possibilities in our research.” Drowning in data Irisson is far from alone in facing this problem. Underwater cameras and environmental sensors mean modern monitoring of oceans, rivers and lakes now generates far more data than researchers can realistically analyse by hand. “Faster and cheaper monitoring means problems like plankton decline, litter accumulation, oil spills and coral degradation can be picked up and acted on sooner.” Gergely Sipos, iMagine Making sense of that flood of information was the goal of iMagine, an EU-funded research collaboration between aquatic scientists and digital experts from 11 European countries. Running from 2022 to 2025, the project developed AI tools to help researchers process an…