World

In India, AI-enabled cameras are sending out tiger alerts in real time

Pinterest LinkedIn Tumblr

Silently padding through the jungle, the tiger slinks between tree trunks and hanging vines, her stripes a seamless veil among the dappled shadows on the forest floor. Hard to spot for a human — harder still if you’re a deer — but not so difficult for artificial intelligence.

Developed by US-based NGO Resolve, TrailGuard AI is an innovative camera trap that is designed to detect specific species and transmit images of them instantly.

While the technology was originally developed to combat poaching — the camera’s first field-test was in a reserve in East Africa in 2018, where Resolve says it led to the arrest of 30 poachersconservationists in India saw potential for its use in managing human-tiger conflict.

TrailGuard uses an advanced vision chip with embedded AI that can recognize up to 10 species — such as tigers, leopards, elephants and humans — and transmit the data in real-time to park rangers via cell phone signal or long-range radio. Because it only recognizes select species, it uses less energy than regular camera traps, so it can stay in the field for more than two years, rather than needing its battery changed every month.

AI of the tiger

Last year, TrailGuard AI deployed 12 cameras in a two-month trial in the Kanha–Pench corridor in Madhya Pradesh, known as India’s “tiger state.” The 3,150-square-kilometer (1,216-square-mile) landscape includes the Pench Tiger Reserve, the Kanha Tiger Reserve, and a forest corridor connecting the two, and is home to over 300 tigers, the largest population in central India. Tigers, which need extensive space to roam, can freely move between the two reserves, which helps the population flourish and aids genetic diversity.

But the tigers aren’t the only ones who live in the forest: it’s also home to around 600,000 people living in 715 villages scattered through the corridor, and there are 2.7 million people living within five-kilometers (3.1 miles) of the tiger conservation landscape – which can create conflict with the big cats.

One of the most common kinds of human-wildlife conflict is tigers killing livestock. For villagers, this can mean the loss of their livelihood, and can lead to “retaliation killings,” which can have a significant impact on the already endangered tiger population.

But TrailGuard AI’s instant transmission of information can protect these communities, says Piyush Yadav, a conservation technology fellow at Resolve. When the camera takes a photo of one of its target species, it sends the image — and information including the location, the time of detection, and the species detected — via email and instant messaging apps to forest rangers.

“We are able to create this early alert system with that real-time data, (so that) the villagers are aware that there is a tiger 300 meters away from their location,” says Yadav. “Based on that, they can react more effectively to this data.”

If a tiger is spotted near a village, forest rangers can then share this information with the community via Whatsapp or Telegram, giving people time to protect themselves and their livestock. In cases where an attack on livestock is unavoidable, the images are also evidence for villagers to claim compensation from the authorities, meaning payment can be processed faster.

This helps the community become more tolerant to living alongside an apex predator, says Himmat Singh Negi, the former director of Kanha Tiger Reserve.

“When we saw for the first time the kind of results, the output given by the technology, it was amazing,” says Negi. “Those who are directly working on the ground, they were really thrilled actually, and they could really save some of these situations where otherwise, something untoward might have taken place.”

There’s a growing need for technology that can ease human-wildlife conflict: globally, human populations around tiger conservation areas increased by 19.5 million people between 2000 and 2020, and in India, 35% of the tiger population lives permanently outside designated reserves.

“This is not only a camera, rather (it’s) a tool for management, because with the use of this technology you would be in a position to save the life of a human being and then the livestock thriving in those areas — and the tiger itself,” Negi adds.

Increasing accuracy

TrailGuard AI was tested in a second trial last year at a tiger reserve in Dudhwa, a 1,310-square-kilometer (560-square-mile) protected area with around 107 tigers roaming between three sanctuaries, where it led to the arrest of four poachers who entered the forest after dark, says Yadav.

The results of the trials at Kanha-Pench and Dudhwa, published in September in the peer-reviewed journal BioScience, found the cameras to have 98.8% accuracy, and marked the first time that an automatic, AI-enabled camera transmitted images of a wild tiger.

While trials have ended, forest staff continue to use the cameras and receive notifications daily.

In the past year, Resolve has upgraded the vision chip in the camera, which it says will increase the accuracy and run faster. The new cameras will be deployed in the Kanha-Pench and Dudhwa reserves in the next few months, as well as West Bengal state, where they will be used in a new trial to manage human-elephant conflict in the area.

The tech is being commercialized and scaled under a spinout company, Nightjar, which aims to produce its first run of 500 units by March 2024. According to Nightjar, it already has pre-orders from companies that manage wildlife habitats.

As apex predators, tigers are vital to maintaining the forest ecosystem, which in turn provides sustenance and livelihoods for hundreds of communities. Yadav hopes that TrailGuard will allow tigers and the local people to thrive in the area.

“The villagers are very well aware that tigers are essential for their own living, their own ecosystem, their children’s future,” says Yadav. “The whole point of the work we do is the coexistence factor — that both the species have to survive.”

This post appeared first on cnn.com