AgriChain: A Farmer-to-Market Intelligence Platform Integrating Satellite Imagery, Deep Learning, and Market Analytics for Precision Agriculture

Abstract

Smallholder farmers in developing agricultural economies continue to face significant losses arising from delayed detection of crop stress, undiagnosed plant diseases, and volatile, opaque market pricing. This paper presents AgriChain, an end-to-end farmer-to-market intelligence platform that unifies remote-sensing analytics, deep-learning-based disease diagnosis, and data-driven market forecasting within a single accessible interface. The system ingests multispectral Sentinel-2 imagery through the Google Earth Engine API to compute the Normalized Difference Vegetation Index (NDVI) for continuous crop-health monitoring, while a convolutional neural network trained on annotated leaf-image datasets performs on-demand disease classification from farmer-submitted photographs. A Long Short-Term Memory (LSTM) network forecasts short-term commodity prices from historical mandi (market) records, and a rule-augmented recommendation engine fuses the outputs of these three subsystems to generate actionable, field-specific guidance covering irrigation scheduling, treatment recommendations, and optimal selling windows. The platform is implemented using a React.js frontend, a Node.js/Express backend, and a PostgreSQL/PostGIS spatial database. Experimental evaluation on a held-out test set of 4,120 leaf images across six disease categories yields a classification accuracy of 94.6%, while the price-forecasting module achieves a mean absolute percentage error of 6.8% over a seven-day horizon. Comparative analysis against three existing approaches demonstrates measurable improvements in detection accuracy and decision latency. The results indicate that tightly coupling satellite-derived agronomic indicators with deep learning and market intelligence can meaningfully narrow the information gap between farm-level decisions and market outcomes.

Introduction

The text presents AgriChain, an intelligent agriculture decision-support platform designed to help farmers improve crop productivity, detect diseases early, and make better selling decisions using artificial intelligence (AI), satellite imagery, and market price forecasting.

Agriculture is a major source of livelihood in developing regions, but farmers often lack timely information about crop health, disease outbreaks, and market price trends. Existing agricultural applications usually address these problems separately, forcing farmers to use multiple tools. AgriChain aims to integrate these functions into a single platform that provides simple, actionable recommendations.

The main objectives of the system are to monitor crop health using satellite data, detect crop diseases from leaf photographs, forecast agricultural market prices, and combine these results into prioritized recommendations for farmers.

The proposed system consists of three major analytical components:

1. Satellite-based crop monitoring: Sentinel-2 satellite imagery is processed using Google Earth Engine (GEE) to calculate the Normalized Difference Vegetation Index (NDVI). This index helps assess vegetation health and identify areas of crop stress. Satellite images are obtained approximately every five days, subject to cloud-free data availability.

2. CNN-based disease detection: A Convolutional Neural Network (CNN) analyzes leaf photographs submitted by farmers to classify five disease categories and a healthy category. The model uses image-processing and deep-learning techniques to identify visible symptoms and provide disease predictions.

3. LSTM-based market price forecasting: A Long Short-Term Memory (LSTM) neural network analyzes historical agricultural market prices to predict short-term price trends. These forecasts help farmers decide when to sell their crops.

The outputs from these three components are combined by a rule-based recommendation engine, which generates clear suggestions based on crop health, disease classification confidence, and expected market prices. For example, the platform can help farmers identify stressed crops, recognize potential diseases, and assess whether selling immediately or waiting for a better price may be appropriate.

AgriChain is developed using React.js for the frontend, Node.js and Express for the backend, PostgreSQL/PostGIS for data storage and geospatial processing, TensorFlow for deep learning, and OpenCV for image processing. It integrates satellite data through Google Earth Engine and uses historical market-price records for forecasting.

The literature review highlights limitations in existing approaches: disease detection is often tested separately from practical farming decisions, satellite monitoring rarely provides direct recommendations, and price forecasting generally ignores crop health. AgriChain addresses these gaps by combining agricultural and economic information in one decision-support system.

The study evaluates disease detection and price forecasting using metrics such as accuracy, precision, recall, F1-score, and confusion matrices. It also recognizes challenges such as poor internet connectivity, variations in field image quality, cloud cover, and the need for reliable predictions under real-world farming conditions.

Conclusion

This paper presented AgriChain, a farmer-to-market intelligence platform that integrates Sentinel-2 satellite imagery, CNN-based crop disease detection, and LSTM-based market price forecasting within a unified recommendation framework. Experimental results demonstrate that the proposed lightweight CNN achieves 94.6% classification accuracy across six disease categories while using substantially fewer parameters than comparable baseline architectures, and that the LSTM-based price forecaster reduces mean absolute percentage error relative to a classical ARIMA baseline. By fusing agronomic and market signals into a single, prioritized set of recommendations, AgriChain addresses a gap identified consistently across the reviewed literature: the absence of platforms that jointly reason over crop health and market timing. The proposed system and its modular architecture provide a practical foundation for future field deployment and longitudinal evaluation with smallholder farming communities.

References

[1] Food and Agriculture Organization of the United Nations, “The State of Food and Agriculture 2023,” FAO, Rome, Italy, 2023. [2] European Space Agency, “Sentinel-2 User Handbook,” ESA Standard Document, 2022. [3] N. Gorelick, M. Hancher, M. Dixon, S. Ilyushchenko, D. Thau, and R. Moore, “Google Earth Engine: Planetary-scale geospatial analysis for everyone,” Remote Sensing of Environment, vol. 202, pp. 18–27, 2017. [4] S. P. Mohanty, D. P. Hughes, and M. Salathe, “Using deep learning for´ image-based plant disease detection,” Frontiers in Plant Science, vol. 7, art. 1419, 2016. [5] K. P. Ferentinos, “Deep learning models for plant disease detection and diagnosis,” Computers and Electronics in Agriculture, vol. 145, pp. 311–318, 2018. [6] E. C. Too, L. Yujian, S. Njuki, and L. Yingchun, “A comparative study of fine-tuning deep learning models for plant disease identification,” Computers and Electronics in Agriculture, vol. 161, pp. 272–279, 2019. [7] J. W. Rouse, R. H. Haas, J. A. Schell, and D. W. Deering, “Monitoring vegetation systems in the Great Plains with ERTS,” in Proc. 3rd Earth Resources Technology Satellite-1 Symposium, NASA SP-351, 1974, pp. 309–317. [8] J. Segarra, M. L. Buchaillot, J. L. Araus, and S. C. Kefauver, “Remote sensing for precision agriculture: Sentinel-2 improved features and applications,” Agronomy, vol. 10, no. 5, art. 641, 2020. [9] P. Thakur, S. Sharma, and P. Kumar, “Vision transformer for plant disease detection: A performance evaluation,” MDPI Sensors, vol. 23, no. 8, art. 4001, 2023. [10] H. Ouyang, X. Wei, and Q. Wu, “Agricultural commodity futures prices prediction via long- and short-term time series network,” Journal of Applied Economics, vol. 22, no. 1, pp. 468–483, 2019. [11] A. Kumar, R. Sharma, and V. Gupta, “LSTM-based agricultural commodity price forecasting for Indian markets,” Springer Applied Intelligence, vol. 52, pp. 12045–12058, 2022. [12] M. Patel, R. Desai, and K. Shah, “AgriApp: A mobile-based crop disease diagnosis system using deep learning,” in Proc. IEEE Int. Conf. on Computing, Communication and Networking Technologies (ICCCNT), 2022, pp. 1–6. [13] L. Zhang, Y. Chen, and H. Wu, “SmartAgri: An IoT-based framework for real-time crop and soil monitoring,” Elsevier Computers and Electronics in Agriculture, vol. 206, art. 107678, 2023. [14] S. Verma and A. Singh, “Artificial intelligence in Indian agriculture: A systematic review and research gaps,” Nature Scientific Reports, vol. 14, art. 3021, 2024.

Copyright

Copyright © 2026 Suraj Mude, Mansi Rahangdale, Kshitij Deshmukh, Shawn Gregory, Prof. Priti Bhagat. 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.