V Ravichandran is a fourth-generation farmer from Poongulam village in Tamil Nadu's Thiruvarur district. The 62-year-old owns 52 acres of land on which he grows rice, cotton, pulses, sugarcane and black gram. For years, pests had been attacking his black gram crop, and it took days, even weeks, to consult agricultural experts. By the time he got the remedies, the infection spread, resulting in crop loss.
Not anymore. Last year, just as he spotted shrunken leaves, he downloaded an artificial intelligence (AI)-driven application on his phone and uploaded photographs of the leaves. The app, Plantix, took minutes to diagnose that the crop had crinkle virus infection and suggested remedies. The disease, if detected early, is easily controllable by tackling aphid, small sap-sucking insects that act as vectors for the virus. "The disease was diagnosed in two minutes. I started remedial measures that afternoon itself. I also used better irrigation methods and harvested 850 kg black gram per acre, all thanks to AI," says Ravichandran. The earlier output used to be 150 kg per acre.
In 2017, Maharashtra, the country's biggest cotton producer, was hit by its worst pest infection on cotton in recent times. More than 50 per cent cotton farmers lost 30-60 per cent crop to pink ball worms. In 2018, the state government approached Wadhwani AI, a Mumbai-based AI research institute. The institute built a smartphone-based AI solution to help farmers identify the infection early. The solution, in the pilot stage, will be rolled out next year.
AI is revolutionising Indian agriculture - which employs almost 50 per cent of the country's total workforce - as tens of startups and technology firms, including big IT ones such as IBM and Microsoft, come up with solutions to age-old problems faced by the agriculture sector
What about small and marginal farmers? "According to agricultural census 2015/16, the average size of a farm holding is as small as 1.08 hectares. This means farm-size information for benefiting small and marginal farmer needs micro-level data at the farm-gate level as well as sophisticated AI technology," says Jayashree Balasubramanian, Director, Communication and Stakeholder Engagement, MS Swaminathan Research Foundation (MSSRF), Chennai.
A lot of farming involves guesswork. "Most decisions, including when to sow, irrigate, add fertilisers, pesticides, nutrients, and when to harvest, anchor on guesswork and instinct," says Ranveer Chandra, Chief Scientist, Azure Global, Microsoft. "However, if a farmer uses his traditional knowledge with data and data-driven information, agriculture can be more productive, cost-efficient and environmentally friendly," he says.
Microsoft, along with International Crops Research Institute for the Semi-Arid Tropics, has developed an AI-based sowing app powered by its Cortana Intelligence Suite that includes machine learning tools. The app sends sowing advisories. It was rolled out in Andhra Pradesh in 2017 and is being tried in other states too. The development of the app involved handling 30 years of climate data with the help of AI for calculating the Moisture Adequacy Index.
Companies are also using deep-learning algorithms to process data captured by drones, sensors or phones to monitor crop and soil health. Azure, for instance, has built an AI-based solution that can provide soil-level data. Instead of a network of sensors, it uses a smartphone's Wi-Fi chipset to beam signals to the ground and detect soil moisture/conductivity. The analysis gives insights into how much water and fertilisers should be used. "Farmers get their phone close to the ground or put it on a tractor or a bicycle and drive around, generating huge data related to their fields. This data is analysed to get insights that can drive decisions," says Chandra. The team is developing the technology further so that small farmers can also afford it.
Weather Forecast/Pest Control
One big area where AI is being used is prediction of weather. "IBM's Watson Decision Platform provides accurate weather data with the help of AI using satellite images of the farm and historical weather data," says Himanshu Goyal, India Business Leader, The Weather Company, IBM. "Last year saw extended monsoon till November. We could tell farmers that rains will be continuous and they must plan for covering their crops or harvest in a certain manner. The predictions helped them prevent short-term losses," he says.
Tata Coffee, one of India's largest integrated coffee cultivators, uses IBM's Watson Decision Platform for Agriculture to receive weather forecasts and information about soil moisture and temperature at coffee estates. IBM has also tied up with Niti Aayog to deploy its AI-based precision agriculture solution in 10 districts of Madhya Pradesh, Gujarat and Maharashtra.
AI tools can also forecast prices that the crop will fetch in the market. For instance, IBM has developed AI-based price solutions for Karnataka for three major tomato-growing districts of Kolar, Chikkaballapur and Belgavi and two maize-producing districts of Davangere and Haveri.
AI tools are also helping farmers in irrigation. Most Indian farmers still rely on rainfall or use flood irrigation. This lowers yields and also leads to huge water wastage. Only 20-25 per cent farmers with access to irrigation use micro-irrigation techniques such as drips or sprinklers. AI tools can tell farmers the exact water needed to grow a crop. For example, Bangalore-based Avanijal Agri Automation has launched an irrigation automation system to help farmers configure and monitor irrigation systems remotely. "AI provides recommendations on how much irrigation should be done a day based on climate and soil moisture," says H.S. Vijayeendra, Co-founder and Director at Avanijal Agri Automation. "This is done by using AI to process satellite images. AI-based irrigation systems also help farmers minimise labour costs. "The in-house AI-based mobile application to control the irrigation system has reduced our labour requirement by 1/6th," says Sudhir Devkar, a hydroponic farmer from Maharashtra and founder of Kryzen Biotech. The company has also developed an AI system to track prices of more than 50 crops in Mumbai, Pune and Bangalore markets daily. "This gives us flexibility to project market trends, requirements and prices to decide crop cycles and expected return," says Devkar.
Pest identification is another area where AI is working wonders. "Earlier, when changes occurred in leaves, we used to make assessments using our experience and intuition," says Ravichandran. "I used to take the leaves to the Krishi Vigyan Kendra more than 80 km from my field. By the time I reached, the leaves would become dry, making it difficult to diagnose the disease. AI and machine learning changed this."
Various states have also developed AI-based advisory solutions for crop monitoring and pest control in local languages. For instance, Tamil Nadu's 'Uzhavan app', released in November 2019, identifies pests and offers remedies in Tamil. A farmer has to upload a photo of the infection on the app. "It has reached more than five lakh farmers," says Santosh K. Misra, CEO of Tamil Nadu e-Governance Agency, which has developed the app.
Most technology firms that offer these tools are not expecting marginal farmers to pay for their solutions. That is why they are working on a B2B model. For instance, CropIn, which was initially a B2C company, realised that the model was expensive for farmers and shifted to the B2B space. "We work with enterprises that have access to farmers, for instance, companies that are into contract farming, fertiliser companies and financial institutions such as banks," says Shah.
IBM works with governments, banks, insurers and companies in the agriculture sector. "Usually, costing for IBMs Watson Decision Platform for Agriculture is calculated based on acreage and portfolio of our weather data services that are subscribed to," says Goyal of IBM.