AI based digital initiatives in Agriculture

The AI implementation strategy adopted by the Ministry of Agriculture and Farmers’ Empowerment (MoA&FW) encompasses a comprehensive approach aimed at harnessing the power of AI to address the unique challenges faced by farmers.

By integrating AI technologies into various aspects of policymaking, resource management, research, and service delivery, the MoA&FW aims to achieve more efficient and effective outcomes.

Kisan e-Mitra chatbot

The Kisan e-Mitra chatbot is powered by AI and leverages speech to text technology and models that identify query intent (based on farmer input)
which then connects to the backend to either provide information or retrieve status and communicate response. 

What does it do

This is an AI-powered chatbot for farmers for grievance redressal, designed to promptly assist farmers with issues and complaints pertaining to payment, registration, eligibility, eKYC updating. The chatbot operates in the farmer's local language, facilitating instant support and streamlined grievance resolution for the central and state governments by automating scheme-related process, thereby reducing manual workload.  

Technical approach of Kisan e-Mitra chatbot

The vision is to make Kisan E-Mitra a one-stop solution for scheme related queries!

Farmer Feedback & Innovation Repository

The farmer innovation repository utilizes artificial intelligence (AI) to capture ground-level farmer practices via multi-modal inputs (voice, text, video). This information captured from farmers is then processed and analyzed to extract insights and identify common themes. These insights are then fed back into the system as peer-vetted practices that are now available across solutions to democratize access to local farmer knowledge.

The solution aims to empower farmers with valuable knowledge and support, ultimately leading to enhanced productivityand sustainability in
agriculture.

VISTAAR - Virtually Integrated System to Access Agricultural Resources

VISTAAR is an innovative open-source digital platform developed to collaborate with multiple partners across the agri-ecosystem, to address challenges in the country's agricultural extension system. It recognizes that no single organization can meet the diverse range of needs and aspirations of India’s vast scale and diversity of farmers. VISTAAR aims to transform advisory services for India's 200 million small-scale farmers. The VISTAAR bot is being used by Frontline Extension Workers to access information based on farmer needs, powered by a library of wide-ranging content to ensure robust information provision.

The Krishi Saathi (Knowledge Management System) is an AI powered chatbot solution that makes use of large language models and language translation models which will be trained using relevant content from verified sources such as Integrated plant management packages, weather information from the Indian Metereological Department, mandi prices from Agmarknet, etc. to support Farmer Tele Advisors in rapid farmer query resolution. 

National Pest Surveillance System (NPSS)

The National Pest Surveillance System (NPSS) is an advanced warning and advisory system utilizing computer vision technology, designed for farmers and agricultural experts to promptly detect and address pests and diseases affecting crops of national importance. This system delivers personalized advisories, curated by experts, offering effective strategies to mitigate the risk of crop loss when implemented.  

The NPSS system's effective implementation aims to enable the timely identification of pest and disease outbreaks. The goal is to issue personalised crop protection advisories in real-time, benefitting 30 million farmers and incorporate more crops to eventually service all pest related grievances.

An Al Enhanced Crop Classification Solution

The solution is a classifier that is powered by XGBoost and SimpleRNN technology, both of which are ML models that process complex data from crop NDVI pattern from Senitel 2 satellites to provide robust crop classification that augment and fill gaps in ground-truthing data to provide refined, vetted information that can then be used to supplement systems of precision agriculture.

Integrated Agricultural News Monitoring System: Krishi 24*7

Krishi 24/7 is a means of integrated agricultural news monitoring that facilitates comprehensive media scanning to capture all agricultural alerts. This will aid the Government on planning and addressing live challenges such as pest attacks, unpredictable weather patterns, and fluctuating market prices throughout India. 

Krishi 24/7 is powered by AI to translate regional news across 12 languages to english, condense and classify breaking news and deliver a personalized feed based on this, tailored to the consumer.

Procedural Transcription Systems for Farmers

This initiative aims to evolve a digital brain through the capture/recording of conversa- tions of FPO meetings for getting Key insights, Key reflections, Call to Actions, Per- spective and discussion summary, Sugges- tions and Questions on various issues discussed in the meetings. Conversations placed across topics and themes across time, aid FPOs to amplify their knowledge over time and improve their performance.

Unified Portal for Agricultural Statistics - AI features

The Unified Portal for Agricultural Statistics (UPAg) is an advanced agricultural data management platform designed to generate crop estimates and integrate with other systems generating Agriculture Statistics such as Price, Trade, Procurement, Stock etc. UPAg aims to empower stakeholders in the agriculture sector, including policymakers, researchers, and farmers, by providing them with comprehensive insights to support informed decision-making.

Al-powered Functions in UPAg

• As the portal grows in capacity and continues to function as a central hub for agricultural data, the MoA&FW plans to leverage Al to enable visualization and forecasting functions to aid landscape analysis and decision-making.
• Predictive Modelling: Al can analyze various data sources across the portal to predict and generate insights on possible future trends. For example, Al can predict trends in market prices, crop production volumes disaggregated by locations, and flag anomalies by providing system alerts that can aid speedy decision-making and support policy efforts
• Robust Data Retrieval: The portal hosts multi-modal content that is currently parsed and viewed in sections: using Al would enable

Digital Crop Survey: Photo Analytics

The digital crop survey has been introduced to provide accurate and correct information on the crops grown by farmers to State and Central Governments to enable seamless benefit delivery.

Process of Digital Crop Survey

• Data is taken from Farmer and farmland data. Geo-referenced map data
• Geo-fences are created to enable control over the location where the crop information is captured
• System mandates capture of photographs along with crop data to ensure accurate information
• Digital Crop Survey provides plot-level crop-sown data for every season with geotagged photographs

Leveraging Al

The Digital Crop Survey currently relies on field-worker/- farmer-captured photographs that are then tagged and cross-referenced to crops. However, the margin for error with respect to crop identification is high given the extent of manual intervention this requires. The MoA&FW is exploring ways in which pictorial analyses can be used to correctly infer crop photographs to eliminate errors that could feed into estimation and modelling.

Assessing Impact & Incorporating Feedback

The conventional approach of conducting assessments often based on measured observations using various experimental designs are mostly expensive, time-consuming and require highly skilled evaluators. These are often associated with human bias involved. In the age of exponential technology like Al, assessments can be readily conducted in a way which is much more convenient, low cost, bias-free and at speed. Such assessments captured in the form of voices/audio responses coming from the targeted farmers are more reliable and trustworthy as they can be traced back to the network of farmers who are the beneficiaries of various schemes and projects.

Al-powered assessments give a complete 360-degree view of the impact that can be synthesized based on the input provided by the targeted beneficiaries. This method of impact assessment is envisioned to cover all schemes under the MoA&FW.

Source : MoA&FW - Compendium of AI and Digital Agriculture