Wildlife officials at one of the most populous tiger zones in India are experimenting with a new set of camera traps — or cameras with infrared sensors deployed in forests to count wild animals or monitor the movement of potential poachers — that harness the power of artificial intelligence (AI).
In experiments under way at the Kanha-Pench corridor in Madhya Pradesh, the system has, for the first time, caught poachers on camera that — a year down — helped forest authorities secure a conviction, according to a person involved in the experiment.
Cameras set up in remote regions are a key tool in surveys and census counts of tigers and elephants but have limitations such as requiring personnel physically access the machine to prise photos. They are also bulky and are not always optimised to conserve charge to keep the instruments running.
The new kind of camera set-up, called the TrailGuard AI camera-alert system, being tested are slim devices that can be inconspicuously set up within the foliage of trees. Shaped like a pen, 13.8 cm long and 1.4 cm wide, it is wired to another ‘communications’ unit, the size of a notepad.
Embedded software
The system has embedded software that can be instructed to take pictures of specific species of interest. In the usual camera set-ups, the motion-triggered devices will snap pictures of anything — from a falling leaf to a jungle fowl that comes in its line of sight. The TrailGuard system can be set to specifically capture humans or species of interest — lions, tigers, cheetahs. If the camera is located at a place within the range of cellphone towers, it can send pictures within 30 seconds. If it is out of this range, it can rely on a longer protocol that can take from 3-10 minutes.
A team of researchers and developers involved in developing and testing the TrailGuard system, in an article published in the peer-reviewed journal Biosciences on September 14 report that 12 TrailGuard AI camera-alert systems were tested simultaneously from mid-May to mid-July 2022 on the Kanha–Pench corridor along with seven camera-alert systems in the Dudhwa Tiger Reserve in early September 2022 to mid-December.
“Researchers and forest department officials received notifications of tiger presence via email or push notification between 30 and 42 seconds after detections. These notifications constitute the first ever transmission of wild tiger detections using embedded AI. These notifications included tiger detections from three units that were within 300 meters of a village and from which there were also daily notifications of villagers grazing cattle or collecting forest products. In addition, a tiger was detected revisiting a recent livestock kill, feeding, and subsequently moving the carcass.,” they note.
Man-animal conflict
India’s tiger population is rising, however a significant proportion of this increase was in zones outside the protected areas. This contributed to rising instances of man-animal conflict which could mean tigers and other carnivores preying on livestock attacked by villagers, attacks on humans as well as poaching. “So far we have never had a technological approach to addressing this issue. But in the Dudhwa [tiger reserve] we learnt that not only could we track [in this case] tigers approaching a village but we also caught a gang of poachers on camera who were walking with a hunted chital. A year down, these poachers were arrested with images from TrailGuard playing a key role,” Piyush Yadav of Nightjar, the company manufacturing TrailGuard units, told The Hindu. The product was developed by RESOLVE, an international non-profit, in which the operative AI-processor is Intel’s Myriad chip. The ‘AI’ element, or ‘embedded AI’ as it is technically known, in this system is that instead of transmitting every image captured, the camera only sends pictures of interest to forest officials. It can thus, if previously programmed so, exclude bears and only send tigers and people. “Transmitting an image is what consumes most of the power and optimising it at the level of the camera means we have been able to have a camera working for nearly a year. In terms of accuracy, it has been able to capture tiger images that have sometimes been missed even by human researchers,” Mr. Yadav added.
Former wildlife officer, H.S. Negi of the Global Tiger Forum, and involved with the project, said the results were “promising” and there were plans to incorporate the system in other tiger reserves.
