SHORT NEWS
Early warning system to save endangered species
Biodiversity is declining rapidly. In order to recognise species worthy of protection in time, a research group from Fribourg wants to combine artificial intelligence, image data and citizen science.
A new road through the forest, droughts or fires: sometimes plants or animals are so decimated within a few years or even weeks that they subsequently appear on the list of endangered species.
Such threats to biodiversity often come surprisingly and rapidly. That is why it is important to be able to sound the alarm at an early stage. This is precisely the goal Daniele Silvestro is pursuing with his SNSF-supported project. In the journal Plants, People, Planet, the researcher from the University of Fribourg outlines an approach that combines artificial intelligence, aerial photography and citizen science. This should make it possible to make the right decisions - and faster than before.
Mobile phone and Citizen Science
The researcher is developing a programme that uses artificial intelligence to evaluate environmental information from various sources such as databases, images and measurements. He wants to optimise it so that it can also analyse satellite and aerial photographs. These images contain a wealth of information: For example, they can reveal deforestation and reforestation, changes in plant cover, new penguin colonies in Antarctica or new infrastructure buildings. "With artificial intelligence, we can analyse millions of images in a very short time," Silvestro explains. "We are thus reaching new dimensions and can monitor the Earth live, so to speak."
To complement the system, Silvestro also wants to incorporate the possibilities of Citizen Science. His vision: volunteers who roam through wasteland, forests or swamps with their mobile phones contribute further images.
Thinking ahead to disaster
With such data, an AI system can not only monitor, but also anticipate problems, identify risk areas and even suggest strategies to prevent environmental disasters. To do this, the team from Fribourg has also further developed a tool that is often used in game apps. "We literally let our AI play. However, we don't want it to defeat an opponent on a chessboard, but to learn strategies that can be used to predict and avert losses in biodiversity," Silvestro explains.