Ridhika Basnet | AUGUST 26, 2026
https://analyticsindiamag.com/ai-features/indias-pest-surveillance-ai-has-reached-national-scale-but-its-impact-is-unproven
Born out of a need to provide farmers with precise crop treatment, India’s National Pest Surveillance System (NPSS) has grown over the last two years into a large-scale digital platform for detecting crop pests and helping farmers.
Launched in August 2024, it has expanded to 31 states and union territories, boasting a user base of nearly 2.5 lakh, including farmers, officials, and extension workers. And its reach is expansive.
The platform has recorded more than 7.92 lakh field surveillance observations and nearly 7.5 lakh pest images, generating more than 36,000 pest and disease advisories. NPSS combines field surveillance with image-based identification and AI-supported analysis.
But what happens after the system identifies a pest? And does the system trace whether the farmers are actually following its advisories?
Inside the NPSS System
NPSS traces back to a 2008-09 outbreak in Maharashtra, when Spodoptera litura (tobacco cutworm or cotton leafworm) and other pests damaged roughly 14.56 lakh hectares of soybean, at an estimated loss of ₹1,392 crore. Also ReadWhy India’s Classrooms Can No Longer Teach the Way They Used To
Niranjan Singh, Principal Scientist at ICAR-National Research Institute for Integrated Pest Management (ICAR-NRIIPM), tells AIM that the episode exposed how “the absence of timely information on pest buildup made it difficult to detect emerging outbreaks and provide timely management advice.” The gap first led to CROPSAP, which has now scaled into a national AI-enabled system.
NPSS has kept its flow simple. Trained scouts, Krishi Vigyan Kendra personnel, and farmers submit pest images or record field observations through a dedicated app or web portal.
The platform’s AI diagnoses pests and diseases, cross-references regional data, and delivers targeted, eco-friendly advisories directly to the user. The aim is to help reduce crop loss while preventing pesticide overuse.
However, the platform doesn’t use proprietary AI. In fact, it’s not even sovereign. The image-identification capability comes from PEAT Inc, the Berlin-based company behind the AI-powered app Plantix that can identify pests across 74 crops and 445 pests. It uses deep neural networks to detect plant diseases, pests, and nutrient deficiencies from photographed symptoms.
“A Plantix/PEAT-provided AI capability integrated into the NPSS application, rather than a model independently developed and trained by the NPSS application team,” Singh notes.Also ReadGujarat Police Seeks AI System to Connect 80,000 CCTV Cameras
The rest is stitched together from other partners. ICAR-NRIIPM built the mobile app and its pest ETL prediction component. The Directorate of Plant Protection, Quarantine and Storage, under the Ministry of Agriculture and Farmer Welfare, supplies technical advisories. Wadhwani AI, a non-profit developing AI solutions, built a model for counting trap-caught pests, particularly in cotton, and BHASHINI translates outputs into regional languages.
Wadhwani AI said that CropAce, its computer-vision model for NPSS, identifies crop pests, diseases and stress conditions from farmer-submitted field images. According to the company, the model is trained on diverse datasets from Indian farms and is optimised for varying lighting, camera quality, and outdoor field conditions. It distinguishes between pest infestations, diseases, nutrient deficiencies, and non-threat conditions, and provides actionable guidance aligned with extension protocols. The aggregated results also help extension teams monitor outbreaks.
What Happens After the Advisory
Identifying a pest is only the first step. If NPSS flags a pest early and an advisory reaches a farmer, the real question is whether that farmer acts on the advisory by using a different pesticide, controlling the volume, or spraying earlier.
What’s your take on this story?Add your comment
However, there is no formal mechanism to track the impact.
“NPSS has so far focused primarily on strengthening early pest detection, surveillance, and timely advisories,” Singh explains. “A formal quantitative assessment of its direct impact on pesticide use, crop losses, or crop yields has not yet been established.” Such an assessment, he adds, “would require systematic field-level comparison and long-term data.” Also ReadHow a 28-Year-Old Product Manager Built an AI Tool to Help Tenants Recover Rental Deposits
Measuring adherence matters, given how steeply compliance drops as pest severity increases. According to ICAR’s own study on NPSS published last year, adherence for pests like yellow stem borer, which infests paddy crops, dropped from nearly 46% in low severity cases to 10% in high severity cases. It noted that while advisory reach was substantial, uptake was lower when recommendations involved costlier or more complex chemical controls.
Any assessment would still need to account for variables beyond the system itself, like weather, pest pressure, and input costs.
Reach
The government reported in February that more than 10,000 extension workers were using NPSS. Beyond the app, the system has been integrated with platforms including agro-advisory platform Krishi Sarthi, Digital Crop Survey, and AI-powered agriculture information network BharatVistaar.
Singh notes that the platform has crossed 2 lakh app downloads and 26,899 registered users, even as its broader base of direct users stands at 2,47,988. That gap, he says, is because “lakhs of users are benefiting from the platform indirectly” through its integration with other tools.
Still Evolving, Still Constrained
The roadmap ahead includes identifying more crops, adding more regional languages beyond the 11 already supported, and building an in-house AI model. But scaling comes with real friction.Also ReadCan Mohali’s Cost Arbitrage Put Punjab on the Global GCC Map?
Singh points to inconsistent internet connectivity in some field locations, variation in the quality of pest images captured by users, and differences in pest appearance across crops and regions, alongside the simpler challenge of encouraging farmers to report pest outbreaks.
The next stage is establishing what the advice actually changes on the ground.
A Larger Trend
NPSS is one part of a much bigger bet on AI in agriculture.
The Digital Agriculture Mission, formed in 2024 to build farmer-centric digital public infrastructure, has a total budget outlay of ₹2,817 crore, including the central share of ₹1,940 crore.
In a reply to a question in the Rajya Sabha, Ramnath Thakur, Union Minister of State for Agriculture and Farmers Welfare, said more than 10.31 crore Farmer IDs were created nationally, and a Digital Crop Survey covered over 31.3 crore plots across 648 districts in Rabi 2025-26. Tools such as NPSS now sit on top of that wider digital base.
What ties these programmes together isn’t just shared infrastructure but a shared gap. In the July 24 reply, the ministry said it “has not undertaken any comprehensive assessment” of the impact of its AI technologies.
India has shown that it can build IDs, surveys, models, and advisories that can reach tens of millions of farmers. What it has not yet shown is whether those systems are changing outcomes in farmers’ fields.