How Artificial Intelligence is helping farmers in Indian villages

TOKYO: Last year, an anguished young farmer in Maharashtra's Amravati district flattened his entire cotton crop spread over three acres of land. What drove him to take the extreme step? His crop was damaged by the dreaded pink bollworm pest, a wily worm that eats away cotton bolls and causes extensive crop damage. Sadly, he is not the only cultivator in the state dealing with the problem.
Thousands of farmers in Maharashtra have been grappling with the pink bollworm menace for several years despite switching to the genetically-altered BT seeds, which are supposed to resist the pest. While experts have identified several reasons behind the menace, lack of timely intervention and shortage of expertise are two major factors adding to the trouble.
Farmers largely depend on government extension workers who manually identify pests by examining pest/insect traps on farms and relay the data to research institutes for expert advice. However, the data is not always reliable since field workers are not equipped with adequate technology to identify or count the pests affecting the crop — this is where Artificial Intelligence (AI) can help fill a critical gap.
A new AI-based pest management tool, which works on a smartphone, is currently being tested in Maharashtra to identify the pink bollworm pest affecting cotton farmers in the state. Developed by Mumbai-based Wadhwani Institute for Artificial Intelligence, the tool classifies and counts pests based on photos taken by a farmer or a field worker on the phone. Wadhwani AI is currently testing this model in partnership with Better Cotton Initiative and the Maharashtra government.
The research institute, inaugurated by Prime Minister Narendra Modi last year, recently bagged a $2 million grant from Google under its AI impact challenge for its pest management solution.
"The model can work on a basic smartphone and there has been incredible acceptance from field workers so far. All they have to do is take pictures of the pest trap on their phone and the system will use image classification models to detect the pest and suggest ways to tackle the problem, such as which pesticides to use, etc," Wadhwani AI's VP (Products and Programmes) Raghu Dharamraju said, adding that the technology can prevent crop damage worth crores of rupees every year.
He said that the approach can be generalized to other crops and the institute plans to find partners to identify them.
Raghu Dharamraju at a Google event in Tokyo
Asked how AI-driven technology can work in rural areas, which are saddled with issues such as poor internet connectivity and low education, Dharamraju revealed that the institute has achieved model compression which makes AI models small enough to be put in a very basic smartphone and functional offline.
But Artificial Intelligence-based tools are not just helping solve agricultural issues in India.
On the healthcare front, for instance, Wadhwani AI is working on a "virtual weighing machine" to fill a critical gap in the maternal and child health in the country.
Dharamraju said the institute has built a smartphone-based technology which allows frontline workers to screen for low birth weight babies in rural homes. With a quick five-second video of a newborn, the AI-powered virtual weighing machine will provide accurate, tamper-proof, geo-tagged measurements like weight, head circumference and other viral parameters on the phone itself.
"There are 1.25 million health workers that go out in the field every day, Now imagine having such a feature in all their smartphones," he said.
Like Wadhwani AI, several institutes and firms are actively working on AI models to tackle key problems in education, healthcare, environment, infrastructure, among others. While the use of AI-based intervention tools is still in its infancy, experts as well as governments are betting big on harnessing their power to build sustainable and long-term social solutions.