Kochi: Six engineering college teachers have patented an AI-powered camera trap technology that will capture, count, and classify wild animals in real time and send live alerts in case of intrusions into human habitats.

The technology offers a breakthrough in wildlife conservation and minimising man-animal conflict, said the faculty members of Rajagiri School of Engineering & Technology in Kochi's Kakkanad.

The patented technology consists of two components: an AI-powered camera trap and a communication system that wirelessly transmits data to the forest officials' computers. The camera trap automatically captures images when it detects the movement of specific animals.

Inside the trap, an AI tool called 'convolutional neural networks' (CNN) zooms in on key features in the photo or video, such as ear notches (for elephants and bears), stripes (for tigers), and coat patterns (for leopards). It then categorizes the images and adds metadata. The trap is connected to the base station (computer) on forest patrol vehicles. As the vehicle nears the camera trap, the data is wirelessly transmitted to the computer.

Binu A, Dr Selvakumar A, Prof Saritha S, Prof Preetha K G, Prof K R Remesh Babu, Dr Sangeetha U. Photo: Special Arrangement
Binu A, Dr Selvakumar A, Prof Saritha S, Prof Preetha K G, Prof K R Remesh Babu, Dr Sangeetha U. Photo: Special Arrangement

"As of now, we have worked with data on elephants and bears from the Forest Department. With more data, the machine can be taught to learn to differentiate between animals within any species," Binu A, assistant professor in the Department of Artificial Intelligence (AI) and Data Science at Rajagiri and one of the four patent holders.

The Forest Department's camera traps are triggered by motion or body heat, capturing images of all animals passing in front of them. Conservation biologists collect this data weekly and manually analyse it. "It consumes a lot of time," said Binu.

Their new technology eliminates the need for manual data analysis, he said. "Our AI-driven system activates the camera only when the animals we are tracking pass by. Once an image or video is captured, it is immediately analysed, classified, and tagged with a unique ID within the camera trap itself," he said.

They have incorporated another technology in the camera trap to seamlessly retrieve the data. Currently, Forest Department officials must manually retrieve data from the camera traps, transfer it to their computer, and return the storage disk to the trap. This requires them to step out of their vehicle and spend time near the trap.

"We’ve integrated a localised wireless technology that automatically transfers the data to the official's computer as soon as they are 20 to 30 meters from the camera trap," said Prof Preetha K G, professor and head of the Department of Computer Science & Engineering at Rajagiri. Prof Preetha, who holds a PhD in wireless networks, said the same communication network integrated into the camera trap can be tweaked to alert people of wild animal intrusions. "This technology offers a much-needed solution to ensure coexistence while safeguarding animals and humans alike," she said.

But for this to work, the camera traps would need to be installed on the fringes of the forest rather than in buffer zones, said Binu. "This way, people would have enough time to respond to the alert," he said.

The four other patentees are Prof Saritha S, an expert in data mining in the Department of Information Technology at Rajagiri, and Prof K R Remesh Babu, an IT expert and principal-in-charge at the Government Engineering College (GEC) - Palakkad at Sreekrishnapuram, and Associate professors Dr Sangeetha U and Dr Selvakumar A of GEC - Palakkad.

They started working on the technology after a nature camp organised by the Forest Department at Vallakkadavu, 6km from Vandiperiyar on the Thekkady-Kottayam route, in April 2018. "The Forest officials told us about their process of tracking and counting animals. We thought of making it easy with AI," said Binu.

How is this tool different
AI-integrated camera traps are an emerging technology in wildlife monitoring, and as of now, there are no widely available commercial models that offer built-in AI capabilities.

There are AI tools such as Google-powered Wildlife Insights, which focuses on species-level identification, rather than distinguishing between animals of the same species. Flukebook is another tool which identifies marine species, such as whales, by their unique tail flukes. Hotspotter uses pattern-matching algorithms to identify and count zebras and giraffes. StripeSpotter, developed by researchers at Princeton University and the University of Illinois at Chicago, analyses the unique stripes of animals to identify and tag them.

These powerful AI tools are, however, not integrated into the camera traps, so the data must still be retrieved manually and uploaded to these software platforms for analysis, said Binu. The ordinary camera traps indiscriminately click warm and moving objects. "The metadata in our camera traps are analysed when it automatically reaches the computer," he said.

Prof Preetha said they have received inquiries from the private players and the government about the technology. "It will provide an efficient and accurate way to monitor wildlife populations, track migration patterns, and assess ecosystem health," she said. Binu said another AI-powered technology for public policing is pending with the Patent Office.

The comments posted here/below/in the given space are not on behalf of Onmanorama. The person posting the comment will be in sole ownership of its responsibility. According to the central government's IT rules, obscene or offensive statement made against a person, religion, community or nation is a punishable offense, and legal action would be taken against people who indulge in such activities.