FarmAssist AI – An Intelligent Multilingual Support for Farmers
DOI:
https://doi.org/10.34293/iejcsa.v4i2.104Keywords:
Agricultural AI, Machine Learning, Crop Recommendation, Plant Disease Detection, Voice Assistance, NLP, Flask, MobileNetV2, Random Forest, Smart AgricultureAbstract
FarmAssist AI is a comprehensive, AI-powered agricultural assistance platform designed to bridge the technological gap between modern artificial intelligence and the Indian farming community. Agriculture in India employs more than 58 percent of the rural workforce, yet the majority of farmers continue to struggle with limited access to timely information, inadequate pest and disease management knowledge, lack of real-time market intelligence, and poor connectivity to government support schemes. This paper presents a full-stack intelligent web application integrating multiple machine learning models, deep learning-based image analysis, rule-based natural language processing, OpenRouter cloud AI, real-time weather intelligence, and a comprehensive market price service into a single unified platform. The system supports twelve functional modules: User Interaction, Regional Language Processing, Speech Processing, Image Analysis, Query Understanding, Weather Intelligence, Crop Recommendation, Disease Prediction, Decision Support, Government Scheme Advisory, Data Management, and a Response Module. A particularly significant feature is the Voice- Based Farmer Assistance capability, enabling speech-to-speech interaction without requiring text literacy — dramatically expanding accessibility across India's farming population. The crop recommendation engine uses a trained Random Forest classifier on seven soil and environmental parameters. The disease detection module employs a MobileNetV2-based CNN achieving approximately 65 percent validation accuracy after five training epochs. The backend is built on Python Flask with CORS-enabled REST APIs, and the frontend is a modern responsive single-page application with dark mode, voice input, and image upload capabilities. Experimental evaluation demonstrates crop prediction latency below 100 ms, NLP response below 200 ms, and disease detection accuracy of 64% using MobileNetV2.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 International Explore Journal of Computer Science and Applications

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

