Drones and supercomputers from Italian ENEA revolutionize the Adelia census with precise algorithms and high-resolution images

In the most remote and inhospitable continent on the planet, Italian science takes a historic step forward. For the first time, artificial intelligence is being used at the South Pole for the automated monitoring of wildlife. ENEA, through an experimental project to census Adélie penguins, has integrated drones, neural networks and supercomputing in the frozen heart of Antarctica, revolutionizing the method used to study biodiversity in the polar zones.
The system, in use at the Mario Zucchelli Base, located in Terra Nova Bay within the boundaries of the Ross Dependency of New Zealand, allowed a high-precision photographic census: 97 percent of the adult specimens and 89 percent of the chicks were correctly identified through automated processing. The very high-resolution georeferenced orthophotographs (1 cm/px) were created using the drones provided to the expedition, capable of operating even below -30°C and at high altitudes so as not to disturb the animals.

“The accuracy is astonishing and demonstrates how technology can be put to the service of biology without impacting the ecosystem,”

explains Riccardo Scipinotti, ENEA researcher and head of the Italian expedition currently working at the Italian-French Concordia base, at over 3.000 metres above sea level.

A supercomputer deciphers drone images

The key element of the project is the CRESCO supercomputer, operating at the ENEA center in Portici, near Naples, which has allowed training a neural network capable of autonomously distinguishing adult penguins from their chicks. To do so, scientists have used over 400 images collected in the previous eight Antarctic campaigns, building a dataset representative of the different environmental and behavioral conditions of the animals.

“The neural network was fed with more than 3.000 specimens photographed from various angles and with different exposures. This made the algorithm robust and adaptable,”

stresses Samuel Pierattini, expert of the Italian National Agency for New Technologies, Energy and Sustainable Economic Development for ICT systems.

“The system analyses images taken at two strategic stages of the austral summer: in mid-November, when the eggs are hatching, and in January, when the chicks are moving freely in the colonies”.

The advantage over manual counting is enormous: time saving, the possibility of covering unreachable areas and no interference with animals. In practice, the entire ecosystem is observed from a distance, with the precision of a digital magnifying glass.

A Flying Lab That Thinks in Real Time

The ambition of the next phase of the project is even more visionary: to integrate directly into the drones a “lean” artificial intelligence, capable of processing data in real time without the need for transmission to computing centers.
This would allow for instantaneous data collection and much faster scientific response, perfect for evolving scenarios such as those related to climate change.

"We are working to develop a computationally light neural network, which can also operate with the limited computing power of drones. The goal is to provide information already processed in the field, drastically reducing analysis times",

Pierattini explains. In the future, the algorithm will be improved to recognize other species, such as emperor penguins, present in colonies near the Zucchelli Base.
This upgrade will make drones into true “flying laboratories”, capable of autonomously carrying out complex missions in extreme environmental conditions, multiplying the opportunities for ecological research and surveillance in the most remote territories on Earth.

A new era for Italian environmental science

The ENEA project is an integral part of the 40th expedition of the National Research Program in Antarctica (PNRA), funded by the Italian Ministry of University and Research.
The scientific coordination is by CNR, while logistics are managed by ENEA and the icebreaker Laura Bassi is entrusted to OGS. A perfect example of synergy between public bodies, advanced technology and environmental protection.

“The use of artificial intelligence in Antarctica represents a turning point for scientific research: it reduces the human impact on ecosystems and increases the precision of detections, paving the way for new models of environmental monitoring”,

declares George Budillon, professor of Physical Oceanography at the University of Naples Parthenope and national coordinator for oceanographic activities of the PNRA.
The experience accumulated in these years could find application in other fragile contexts of the planet: from the Amazon forests to coral reefs, where non-invasive observation is the key to studying and protecting biodiversity.

An algorithm that finally whispers to penguins

The image of a silent drone flying over a penguin colony, while an AI counts the individuals in real time, may seem like science fiction. Yet, it is already reality. Antarctica, one of the most difficult and delicate territories to study, is becoming an advanced laboratory for testing tools that could soon be essential in other threatened ecosystems.
The ENEA project demonstrates how innovation, respect for nature and scientific collaboration can generate sustainable and effective solutions. And while the penguins continue their survival ritual in the ice, an algorithm will watch over their future.

Here are three insights that might interest you:

Protecting Antarctic waters and the future: Krill research
Bouvet, the remote island home to a neglected web extension
World's Largest Iceberg Stood in South Georgia