An AI-Driven Decision Support System for Personalised Crop Advisory Using Agentic AI, Generative AI, and CNN-Based Disease Detection for Local Farmers
Main Article Content
Abstract
— A large portion of local farmers in India rely on traditional practices and local intermediaries for critical decisions concerning crop selection, pest management, and fertilizer application, often leading to suboptimal yields, higher input costs, and environmental degradation.
This work proposes an AI-driven Decision Support System integrating Agentic AI and Generative AI (GenAI) with Convolutional Neural Networks (CNNs) and Gemini Vision to enable plant disease detection, soil-informed fertiliser recommendation, and multilingual advisory. The proposed system introduces an agentic decision pipeline wherein leaf images are processed concurrently by a fine-tuned CNN for high-precision classification and Gemini Vision for contextual visual reasoning. The outputs are fused by an AI agent to generate robust crop health advisories that combine quantitative accuracy with explainable insights. The system further incorporates real-time, location-aware, and voice-enabled interaction in regional languages, including Hindi, Rajasthani, and Punjabi. Experimental results demonstrate a classification accuracy of 98.3% for disease detection and 94.7% accuracy for fertilizer recommendations against laboratory benchmarks. The agentic framework achieves a 95.5% task completion rate with an average latency of 4.1 seconds, significantly enhancing decision efficiency for farmers
Article Details
Section
COPYRIGHT
Submission of a manuscript implies: that the work described has not been published before, that it is not under consideration for publication elsewhere; that if and when the manuscript is accepted for publication, the authors agree to automatic transfer of the copyright to the publisher.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work
- The journal allows the author(s) to retain publishing rights without restrictions.
- The journal allows the author(s) to hold the copyright without restrictions.
References
[1] NABARD, “All India Rural Financial Inclusion Survey 2021–22,” Na- tional Bank for Agriculture and Rural Development, Mumbai, India, 2022.
[2] R. Bhavana and S. Kumar, “Impact of ICT-based crop advisory services on yield improvement among smallholder farmers in India,” Journal of Agricultural Extension, vol. 24, no. 3, pp. 112–121, 2021.
[3] L. G. Goswami, C. Kavadia, R. Sharma, L. Khandelwal, and M. Sagotia, “Potato plant disease detection using CNN,” in Proc. 3rd Int. Conf. on Multi-Disciplinary Application and Research Technologies (ICMART- 2024), Udaipur, India, May 2024, pp. 252–260.
[4] S. B. Jadhav et al., “Utilizing CNN transfer learning for soybean disease classification: A comparative study,” Journal of Agricultural Informatics, vol. 9, no. 1, pp. 17–26, 2018.
[5] J. Liu et al., “Deep learning approaches for plant disease and pest detection: A comprehensive review,” Plant Disease Detection Journal, vol. 15, no. 2, pp. 45–56, 2019.
[6] D. Zhang et al., “DENS-INCEP: A deep transfer learning approach for rice disease detection,” International Journal of Agricultural Engineer- ing and Technology, vol. 13, no. 4, pp. 112–124, 2020.
[7] J. A. Pandian et al., “A novel deep learning model for plant leaf disease identification using 14-DCNN,” Computers and Electronics in Agriculture, vol. 195, 2022.
[8] P. Bedi et al., “Plant disease detection using hybrid CNN-CAE model,”
Computers in Biology and Medicine, vol. 139, 2021.
[9] N. Rashmi et al., “Machine learning methods for plant disease identifica- tion: A comprehensive analysis,” International Journal of Plant Science and Agricultural Technology, vol. 5, no. 2, pp. 32–41, 2021.
[10] A. Sharma and R. Patel, “Machine learning-based soil nutrient recom- mendation for sustainable crop production,” Computers and Electronics in Agriculture, vol. 185, pp. 106–117, 2021.
[11] S. Kumar et al., “A comprehensive survey of plant leaf diseases classi- fication utilising image processing techniques,” International Journal of Computer Applications, vol. 122, no. 4, pp. 11–16, 2015.
[12] S. Yao et al., “ReAct: Synergizing reasoning and acting in language models,” in Proc. ICLR, 2023.
[13] H. Chen et al., “AgriGPT: Coupling large language models with domain knowledge for agriculture advisory,” arXiv preprint arXiv:2311.09259, 2023.
[14] Google DeepMind, “Gemini: A family of highly capable multimodal models,” Technical Report, 2023.
[15] ICAR, “Soil Health Management: Soil Testing and Fertiliser Recom- mendation Guidelines,” Indian Council of Agricultural Research, New Delhi, 2020.