SmartKrishi-AI: A Survey of AI-Integrated Agriculture Decision Support and Farm Resource Management

Abstract

Agricultural choices depend heavily on a complex mix of environmental variables, including soil properties, fluctuating weather patterns, crop compatibility, market dynamics, pest threats, and the availability of vital farming resources. Most contemporary digital services target only one of these factors, forcing growers to manually gather data from multiple disconnected websites and local contacts. This paper reviews recent developments related to SmartKrishi–AI, a conceptual, integrated platform designed for agricultural decision-making and asset coordination. Our analysis evaluates several vital domains, including crop recommendation frameworks, weather forecasting networks, commodity price models, plant disease diagnostics, public welfare discovery, equipment leasing, and labor scheduling. We analyze recent research from 2023 to 2026 indexed across major venues such as IEEE, Elsevier/ScienceDirect, and MDPI, exploring their mathematical approaches, dataset constraints, evaluation methods, and practical challenges. Based on these insights, we introduce a conceptual multi-layer architecture that connects data-driven analytical models with verified institutional channels and peer-to-peer resource exchange workflows. Because the proposed ecosystem serves as a structural blueprint rather than an experimentally field-tested software tool, no project-specific crop yield or predictive precision increases are claimed. Finally, we highlight persistent open challenges in this sector, such as data quality gaps, geographic adaptation limits, model interpretability, forecast volatility, system interoperability, regional language support, and the necessity for rigorous field trials.

Introduction

The text presents SmartKrishi–AI, a conceptual AI-powered smart agriculture platform designed to help farmers make better decisions throughout the entire crop cycle, from crop selection and cultivation to harvesting and resource management.

Traditional agricultural decisions often depend on fragmented information, such as separate soil-testing reports, weather forecasts, market prices and disease alerts. SmartKrishi–AI aims to integrate these different sources into one unified platform using Artificial Intelligence (AI), Machine Learning (ML), Deep Learning (DL), weather analytics and digital marketplaces.

1. Introduction and Problem Statement

Farmers face several challenges when making agricultural decisions:

• Selecting suitable crops based on soil conditions and climate.

• Predicting weather conditions and seasonal risks.

• Identifying crop diseases and pest attacks.

• Forecasting future market prices.

• Finding suitable agricultural machinery and labour.

• Discovering government agricultural schemes and subsidies.

Existing solutions generally address these problems separately, making it difficult for farmers to obtain complete and timely information.

To overcome these limitations, SmartKrishi–AI proposes an integrated decision-support platform that combines agricultural intelligence, real-time information and resource-sharing facilities.

2. Farmer Requirements

The platform focuses on supporting farmers throughout four major agricultural stages:

• Pre-sowing: Soil analysis, crop selection, budget planning and seasonal assessment.

• Cultivation: Weather monitoring, irrigation guidance and disease detection.

• Harvesting: Market price forecasting, harvesting schedules and storage planning.

• Post-harvesting: Machinery availability, labour hiring and resource management.

The platform also aims to provide regional-language support, explainable recommendations, reliable information sources, data security and mobile accessibility.

3. Recent AI Approaches in Smart Agriculture

A. Crop Recommendation and Explainable AI

Machine learning algorithms such as Decision Trees, Random Forest, Support Vector Machines (SVM) and Gradient Boosting can recommend suitable crops based on soil nutrients (NPK), pH, temperature, humidity and rainfall.

Explainable AI techniques such as LIME help farmers understand why a particular crop is recommended.

B. Weather Intelligence

Weather forecasting and deep learning models help predict rainfall, temperature and other environmental conditions. The platform integrates official weather forecasts with localized weather analysis to support irrigation, crop planning and risk management.

C. Market Price Forecasting

Machine learning and deep learning models analyse historical market prices, seasonal trends, supply and demand to estimate future commodity prices.

The system presents expected price ranges and prediction uncertainty rather than guaranteeing a particular selling price.

D. Crop Disease and Pest Detection

Computer vision and deep learning models analyse crop images to identify possible diseases and pest-related symptoms.

The system provides suspected disease information, confidence levels and visual evidence. Farmers are encouraged to consult agricultural experts before taking treatment-related decisions.

E. Equipment Rental and Labour Hiring

The platform provides digital directories where farmers can find and rent agricultural machinery or hire labour.

Features include:

• Machinery details and rental prices.

• Equipment owner verification.

• Location-based availability.

• Labour skills and availability.

• Booking and scheduling facilities.

F. Government Schemes and Advisory Information

The platform helps farmers discover government agricultural schemes, subsidies, eligibility requirements, application deadlines and official portals.

All eligibility information is treated as preliminary guidance and must be verified through official government sources.

4. Data and Intelligence Pipeline

The proposed platform collects information from multiple sources, including:

• Farmer details and geographical location.

• Soil test reports and agricultural land information.

• Weather forecasts and historical climate data.

• Historical agricultural market prices.

• Crop images for disease identification.

• Machinery and labour availability records.

Different AI models process these inputs to generate recommendations.

The platform compares several machine learning algorithms and evaluates their performance before selecting suitable models. It also maintains data quality, source verification and geographical validation to improve reliability.

5. Benchmark Results and Research Analysis

The research discusses an existing crop recommendation study by Tomar et al. that evaluated multiple machine learning algorithms using soil and climate data.

These results demonstrate the potential of machine learning for crop classification. However, the study highlights that high accuracy on public datasets does not necessarily guarantee similar performance under real agricultural conditions.

Therefore, SmartKrishi–AI proposes additional evaluation using macro-F1, precision, recall, probability calibration and geographical validation.

6. Identified Research Gaps

The research identifies four major limitations in existing agricultural technologies:

1. Fragmented Functionality: Crop recommendation, weather forecasting, disease detection, market prediction and resource management are generally available as separate systems.

2. Local Generalization and Data Quality: Models trained on limited or generic datasets may not perform accurately across different geographical regions, soil conditions and seasons.

3. Explainability and Uncertainty: Existing systems often provide fixed predictions without explaining the reasons or uncertainty behind their recommendations.

4. Accessibility and Governance: Farmers face challenges related to internet connectivity, digital literacy, regional languages, data privacy and access to reliable agricultural experts.

SmartKrishi–AI aims to address these gaps through a unified and farmer-centric architecture.

7. Proposed SmartKrishi–AI Architecture

The proposed system is a modular web and mobile platform with five main user roles:

• Farmer

• Equipment Provider

• Labour Provider

• Agronomic Expert

• System Administrator

The architecture consists of seven functional layers:

1. User Interface Layer: Provides multilingual dashboards, forms, notifications and search facilities.

2. Application and API Layer: Handles authentication, user permissions and communication between modules.

3. Intelligence Processing Layer: Executes AI models for crop recommendation, weather analysis, market forecasting and disease detection.

4. Verified Information Layer: Manages government schemes, agricultural advisories and expert verification.

5. Resource Exchange Layer: Supports machinery rental, labour hiring and booking management.

6. Data Persistence Layer: Stores user profiles, soil records, historical data, transactions and model results.

7. Monitoring and MLOps Layer: Monitors model performance, system errors, data changes and user feedback.

Technology architecture:

• Frontend: Web and mobile interfaces.

• Backend: Java with Spring Boot.

• AI and ML services: Python.

• Database: Secure relational database.

• External integrations: Weather APIs, market data feeds and government portals.

8. Evaluation Methodology

The research proposes evaluating the platform at three levels.

A. AI Model Evaluation

• Crop recommendation: Accuracy, precision, recall and macro-F1.

• Market forecasting: MAE, RMSE and MAPE.

• Weather prediction: Forecast error across different locations and time periods.

• Disease detection: Precision, recall, F1-score and confidence calibration.

B. System Evaluation

• Response time and system availability.

• API performance and data freshness.

• Error rates and outdated marketplace listings.

• Availability of verified government scheme information.

C. User-Level Evaluation

• Machinery booking success rate.

• Equipment verification time.

• User satisfaction and trust.

• Accessibility and usability across different agricultural regions.

Real-world testing should involve farmers from multiple geographical regions with proper consent.

Conclusion

This comprehensive survey evaluated recent peer-reviewed literature relevant to the design of SmartKrishi–AI, a unified agricultural decision-enrichment and resource-management architecture. Contemporary studies [1]–[4], [13], [15]–[18] outline valuable algorithmic progress across machine-learning-driven crop selection, model interpretability, and IoT precision agriculture networks. Concurrently, specialized investigations [5]–[9] provide robust foundations for multi-market price modeling, weather downscaling, and computer-vision plant health monitoring. The primary operational opportunity lies in connecting these separate analytical blocks with verified public program registries and practical, localized machinery and labor exchange networks [14]. However, the proposed architecture remains entirely conceptual and demands careful software engineering, extensive regional validation, localized usability studies, and rigorous multi-metric evaluation before any empirical claims regarding yield optimization or farmer economic security can be substantiated.

References

[1] “Machine Learning Based Smart Crop Recommendation System,” in Proc. 2nd Int. Conf. Computational Intelligence and Computing Applications (ICCICA), 2025, doi: 10.1109/ICCICA67008.2025.11337597. [2] “A Data-Driven Crop Recommendation System with Explainable AI for Precision Agriculture,” in Proc. 3rd Int. Conf. Augmented Intelligence and Sustainable Systems (ICAISS), 2025, doi: 10.1109/ICAISS61471.2025.11041903. [3] “Optimizing Crop Recommendation with IoT and Machine Learning: A Smart Agriculture Framework 4.0,” in Proc. IEEE 2nd Int. Conf. Computing, Applications and Systems (COMPAS), 2025, doi: 10.1109/COMPAS67506.2025.11381806. [4] “Intelligent Crop Recommendation Framework for Optimized Agricultural Productivity using Machine Learning Techniques,” in Proc. Int. Conf. Electronics and Renewable Systems (ICEARS), 2026, doi: 10.1109/ICEARS67481.2026.11416759. [5] B. A. Celik and S. Celik, “Hybrid forecasting of agricultural commodity prices: Integrating machine learning, time series, and stochastic simulation models,” Borsa Istanbul Rev., vol. 25, no. 6, pp. 1440–1462, Nov. 2025, doi: 10.1016/j.bir.2025.10.004. [6] P.-Y. Kow et al., “AI-driven weather downscaling for smart agriculture using autoencoders and transformers,” Comput. Electron. Agric., vol. 232, Art. no. 110129, May 2025, doi: 10.1016/j.compag.2025.110129. [7] “An integrated framework for multi-commodity agricultural price forecasting and anomaly detection using attention-boosted models,” J. Agric. Food Res., vol. 22, Art. no. 102021, Aug. 2025, doi: 10.1016/j.jafr.2025.102021. [8] X. Liu et al., “Solutions and challenges in AI-based pest and disease recognition,” Comput. Electron. Agric., vol. 238, Art. no. 110775, Nov. 2025, doi: 10.1016/j.compag.2025.110775. [9] J. Yan et al., “Recent developments and applications of crop disease detection, prediction, and early warning: A review,” Engineering, vol. 62, pp. 316–340, Jul. 2026, doi: 10.1016/j.eng.2025.10.032. [10] “AI-Driven Crop and Fertilizer Recommendation: A Multi-Step Predictive Modelling Approach,” in Proc. IEEE 3rd Int. Symp. Sustainable Energy, Signal Processing and Cybersecurity (iSSSC), 2025, doi: 10.1109/iSSSC66652.2025.11388308. [11] “Crop Recommendation System for Best Farming Practices Using Various Deep Learning Algorithms,” in Proc. Int. Conf. Emerging Technologies in Engineering Applications (ICETEA), 2025, doi: 10.1109/ICETEA64585.2025.11100010. [12] Food and Agriculture Organization of the United Nations, “Digital technologies in agriculture and rural areas: Status report,” FAO, Rome, Italy, 2019. [13] V. Tomar, G. Sharma, R. Rajkumar, A. Singh, S. S. Dhakare, and K. Kumar, “Development of a machine learning-based system for optimizing crop recommendations,” in Proc. 1st Int. Conf. Advances in Computing, Communication and Networking (ICAC2N), 2024, pp. 1594–1599, doi: 10.1109/ICAC2N63387.2024.10895212. [14] G. Rawat, S. Faisal, M. Asim, and N. Bansal, “Empowering Indian farmers through agri-equipment sharing: A sustainable approach for small-scale farmers,” in Proc. 1st Int. Conf. Advances in Computing, Communication and Networking (ICAC2N), 2024, pp. 458–463, doi: 10.1109/ICAC2N63387.2024.10895182. [15] M. K. Senapaty, A. Ray, and N. Padhy, “A decision support system for crop recommendation using machine learning classification algorithms,” Agriculture, vol. 14, no. 8, p. 1256, Jul. 2024, doi: 10.3390/agriculture14081256. [16] H. Lahza et al., “Optimization of crop recommendations using novel machine learning techniques,” Sustainability, vol. 15, no. 11, p. 8836, May 2023, doi: 10.3390/su15118836. [17] C. Musanase, A. Vodacek, D. Hanyurwimfura, A. Uwitonze, and I. Kabandana, “Data-driven analysis and machine learning-based crop and fertilizer recommendation system for revolutionizing farming practices,” Agriculture, vol. 13, no. 11, p. 2141, Nov. 2023, doi: 10.3390/agriculture13112141. [18] E. Elbasi et al., “Crop prediction model using machine learning algorithms,” Appl. Sci., vol. 13, no. 16, p. 9288, Aug. 2023, doi: 10.3390/app13169288.

Copyright

Copyright © 2026 Namrata Shinde, Tejashree Torave, Gayatri Bhise, Suhani Yadav, Prof. P. N. Kadam. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.