Nature conservation 2.0

50 elephants killed by trains: AI to save animals

Archive image: This elephant survived a collision with a train in India - with extensive abrasions. But many other elephants are less fortunate.

(Bild: APA/AFP/STR)

When elephants move around India in search of food, they often cross railroad tracks that cut through their habitats. Collisions with trains are the second most common unnatural cause of death for elephants there. According to official figures, at least 50 animals have died in this way in the past four years. Monitoring and warning systems with AI are now intended to better protect animals on the subcontinent.

The Indian Railways, for example, is currently installing technology and cables along tracks in the north-east, which can detect vibrations caused by the footsteps of elephants within a five-meter radius. Notifications are then sent to a mobile app and a control room. Approaching trains can thus be automatically warned to slow down or stop. A different system is being tested in the south of the country.

AI can not only help in India
Artificial intelligence for animal welfare - ideas like this are not just happening in India, but in many countries around the world, says ecologist Arnulf Köhncke, Head of Species Conservation at the environmental organization WWF Germany. AI is used particularly frequently to evaluate images, explains the expert on the occasion of World Wildlife Day on October 4. "Because we want to know how many animals of a species are living somewhere." This requires setting up camera traps, counting animals in the photos and statistically evaluating the data - "without support, it takes a long time".

This is because the cameras take countless photos. "The AI helps to determine what is an animal in the photos and what is not." This allows the huge amounts of data to be handled better, says Köhncke. The AI can also recognize which species can be seen in the photos - and sometimes even which individual animals. "In the case of cats such as tigers and leopards, it can recognize the individuals by their stripes and spots."

A tiger's markings are the animal's "fingerprint". AI can use this knowledge to identify individuals.

(Bild: belizar - stock.adobe.com)

Even in zebras, giraffes, whales and dolphins, patterns or fins are unique and can therefore be decoded. "If you can recognize the animals individually, you can use statistical models to calculate how many animals of this species there are in total in the area," explains the expert.

WWF and IBM count African elephants with AI
WWF Germany launched such a project in August together with IBM in Central Africa, where forest elephants are to be observed and counted. AI not only helps to analyze the data volumes in such projects. It can also recognize behavioural patterns and make precise predictions.

Other AI approaches are audio-based, continues Köhncke. This makes a lot of sense in large forest systems in Africa, Asia or Latin America in particular, because you can't see very far there. Some programs can recognize bird calls and thus help to map animal species. Experts can even draw conclusions from chainsaw noises - namely where forests are being cut down. And gunshots indicate poachers.

Satellite photos can also be analyzed
Monitoring can also be carried out from space. Köhncke explains that satellite photos can quickly show where there is a fire. "Satellite analysis also makes it possible to record the movement of vultures." This makes it possible to identify where a dead animal is lying - which could also be an indication of poaching. "More and more things are being developed," summarizes Köhncke.

The Leibniz Center for Tropical Marine Research in Bremen recently reported on using AI to look into the depths. Two species of surgeonfish were observed in coral reefs in the Red Sea. The movements of the fish were recorded in three dimensions while they were searching for food. This enabled the researchers to better understand the marine ecosystem - which is important for developing protective measures for the reefs.

In India, experts are also thinking about how to prevent encounters between humans and potentially dangerous wild animals. After all, people die time and again when they encounter elephants, tigers or leopards. The Chief Conservator of Forests in Uttarakhand in northern India, Sameer Sinha, tells dpa that camera traps with AI technology are now being used in his state for this reason.

The cameras have internet capability, allowing images to be transmitted to a computer server in real time. If such animals come close to villages, warnings are automatically generated. The forestry authority then informs the villagers and deploys response teams to intervene. The only problem with the application so far, says Sinha, is that there is not always good internet in the area.