“Last year, we lost two of our men to poachers,” Samrat Gowda, deputy director of the Similipal Tiger Reserve in Odisha, said. “Every time we come across them, the poachers are armed with a loaded gun.”
But such encounters have become much less common of late. As part of an early alert system called TrailGuard AI, the Similipal Tiger Reserve was fitted with 100-150 cameras loaded with an artificial intelligence (AI) model. The cameras relay images of people and wildlife entering the forest to the model, which looks for the presence of poachers among them.
“Earlier, we didn’t know when poachers entered. Now we have clear information about the area the poachers are in, so our people are prepared,” Gowda said.
In the last 10 months, TrailGuard AI has helped wildlife officials at Similipal arrest 96 poachers and seize more than 86 country-made guns. In December alone, the team arrested over 40 poachers.
“House raids based on photo identification have given very good results,” according to Gowda. “If this trend continues, I’m hopeful poaching can be reduced by at least 80%. Once that happens, naturally, our people will be safer, along with the forest and wildlife.”
Proactive enforcement
The AI-enabled cameras are tucked away in the reserve’s thick vegetation. They operate on a low-power mode by default but switch to a high-power mode when they sense movement, and capture an image. The camera then performs AI inference on the edge, meaning it uses the chip inside to sort between various object classes such as ‘animals’, ‘humans’, and ‘vehicles’ in the image. If the AI deems it necessary, it autonomously transmits an image using the cellular system attached to the camera to an end-user in 30-40 seconds.
“We have set up a control room in our headquarters, with a big screen, where we are alerted whenever there is a photo update,” Gowda said. “We then immediately transmit the information on our WhatsApp groups and VHF radio.”
Catching poachers in the forest still isn’t straightforward. Wildlife officials use intelligence sources to identify the poachers caught on camera. These sources include their regular staff, who go undercover with poachers to collect information about who they are, the villages they hail from, and other details.
“Once we get 100% confirmation that these are the people that entered the forest, we will raid their house or village and arrest the person and forward them to the court with proper documentation,” according to Gowda.
He also stressed the importance of proactive enforcement mechanisms backing up the inputs provided by TrailGuard. “Getting photos is the easy part, but after that what you do is most important. We are actively going and raiding [houses] and bringing people in. So both the technology and our on-ground efforts complement each other to give us good results,” he added.
According to the latest department report, the arrests in 2024 led to one conviction, which was obtained within six months — fast, according to Gowda. His colleagues are hoping for two or three more convictions soon.
Smaller, cheaper, durable
TrailGuard AI was conceived and made by Nightjar Technologies, a social impact enterprise in Gurgaon that develops remote surveillance devices for conservation settings. Its founder Piyush Yadav identified the design of TrailGuard camera systems to be what makes it unique. “There are two units,” he explained. “One is the camera unit, the size of a pen, and the other is the battery/communication unit, the size of a notepad. They are attached using a two-meter-long cable. So it’s not bulky but is rather a spread out design.”
Gowda said the smaller footprint of the device reduces the chances of poachers stealing them.
But according to him, TrailGuard’s best feature is its battery life. “There are so many other technologies available for live transmission, but the [TrailGuard] battery lasts for six months to one year based on the number of photos it sends,” he said. “We don’t have to go in and change the battery again and again.”
This amounts to a blessing in the challenging terrain of Similipal.
“They are not very costly compared to other technologies,” he added. According to Yadav, TrailGuard AI cameras cost roughly Rs 50,000-53,000 per unit.
Access to tribal communities
The villages in and around Similipal are occupied by tribal communities. Hunting is part of their culture, even if many of them have moved away to other forms of sustenance. They need to be able to access the forest, too.
“The tiger reserve has traditionally had a lot of incursions from neighbouring communities around Similipal,” said Aditya Panda, a naturalist, wildlife conservationist, and the honorary wildlife warden of Satkosia Tiger Reserve in the same State. “People come in large numbers to engage in bushmeat poaching.”
Apart from TrailGuard cameras, the forest department uses regular camera traps in areas with no network, rendering it almost impossible for anyone to go into the forest without being caught on camera. (A camera trap is a camera rigged to capture an image every time it senses motion nearby. They are often used to photograph animals in the wild.)
But one result of this surveillance is that many villagers have simply stopped going into the forest: they don’t want their faces to be seen on camera, be mistaken for poachers, and arrested. Their ability to collect firewood and other non-timber forest products has concomitantly declined.
“We are discussing with the [local] people and are facilitating safer ways to access the forest, because it shouldn’t be that because of one poacher everyone is restricted,” Gowda said.
The department is also conducting regular awareness meetings about measures to prevent poaching with the tribal communities in the local language.
Wider use-case, adoption
The TrailGuard system has been an effective anti-poaching tool and Panda believes it can do more. “I think this sort of technology can be a gamechanger when it comes to patrolling and monitoring our protected areas, not only to intercept illegal entry and incursions but also in monitoring wildlife,” he said.
Gowda agreed. He said the department has already profiled most tuskers in the region with the cameras’ help and expressed belief the technology can help ameliorate the local human-wildlife conflicts as well. He added that this avatar of the system will soon be deployed in other parts of the State.
Similipal is the first reserve where TrailGuard has shown success as an ‘anti-poaching tool’ but it has takers outside Odisha as well. “We have deployment ongoing in five States right now, at more than 14 sites,” Yadav said. It has also been implemented in the Kanha Tiger Reserve in Madhya Pradesh and Dudhwa National Park in Uttar Pradesh with 20 and 10 cameras respectively, as part of efforts to mitigate human-wildlife conflict.
But Yadav also said NightJar is trying to keep from scaling up too fast. “It is a complex hardware product. We want to take it step by step, optimise for issues along the way, and be careful with expansion,” he said.
Nikhil Sreekandan is an independent journalist.
Published – February 17, 2025 05:30 am IST