Agriculture is already a tough vocation, but for a group of farmers in Odisha it had become even more challenging due to an unlikely visitor to their fields. Large herds of elephants had been frequently paying visits and damaging crops and the farmers had tried everything conventional, such as fencing, planting lemon saplings and so on, but could not stop the incursions.
That’s when Ammachi Labs, the innovation hub of Amrita Vishwa Vidyapeetham college in Amritapuri, Kollam, came to the farmers’ aid by devising an artificial intelligence (AI)-based tool to solve the human-elephant conflict. Nabard facilitated the connect.
Named ‘Amrita Elephant Watch’, the tool is essentially a video surveillance system embedded with a layer of AI to recognise and monitor elephant movements. Linked to mobile networks, the system sends real-time information of elephant movements to registered mobile numbers via a simple SMS within 15-20 seconds of spotting. Alerts have also been configured to be delivered on the Telegram app, after which farmers can take appropriate action to chase away the elephants and minimise damage.
Solar panels power the cameras, and the issue of darkness is addressed by the camera’s infrared capabilities to recognise movements. While internet connectivity is improving in remote areas, the Ammachi Labs researchers also use private wireless networks such as LoRaWAN (Long-Range, Wide Area Network) to power the alerts from Elephant Watch.
“We have been developing the solution for almost a year now, and recently showcased the prototype at the Government of India’s Global Partnership on AI (GPAI) Summit,” Balu Mohandas Menon, Team Lead, Rural technologies, Ammachi Labs, said. “The tool has now been deployed for three months in the Madukkarai forest (in Coimbatore, Tamil Nadu) and we have had 2-3 instances of elephant alerts including at nights,” he adds.
The Rural Technologies team takes technologies beyond labs to end users in rural areas, and comprises researchers along with student interns from computer science, electronics and other streams.
Sustainable Coexistence
With human-animal conflicts on the rise, some forest departments have also invested in such tools to monitor wildlife movements near railway tracks and other key places. Interestingly, the Tamil Nadu Forest Department launched an AI and ML-enabled surveillance system earlier this year to prevent elephant deaths on railway tracks, and deployed it at the same Madukkarai area. But unlike Ammachi Labs’ simpler solution for the hinterlands, these involve large investments in funds and manpower to control rooms.
The successful sightings and alerts of the tool today comes after a fair share of challenges.
“Any AI system needs to be fed large amounts of training data for it to learn the target. We had to feed the system with over 1.5 lakh elephant images — with tusks, without tusks, day time, night time and others — for it to start recognising the animal on field,” Mohandas said. There were false alarms too. “In the initial days, pigs used to be tagged as elephants since from the top angle, the rear side of a pig resembles an elephant,” he recalls.
The team is also clear that they want to strive for ‘sustainable coexistence with wildlife with this tool’. By design, they want to just facilitate alerts and let the user solve it in a harmonious manner. “We have set up the tools at perimeter locations that are sensitive to humans but at the same time not intruding into areas that offer animals their basic needs,” says Mohandas.