Intelligent Farmer Assistant And Crop Lifecycle Management Platform With Multilingual Support

Abstract

Agriculture supports the livelihoods of more than 140 million households in India, yet a significant disconnect persists between the information farmers require and what is accessible to them. Existing digital advisory tools are often fragmented, linguistically restrictive, and dependent on high levels of technical proficiency, limiting their usability among smallholder and marginal farmers. These barriers are particularly evident in rural regions, where language diversity, inconsistent internet connectivity, and limited digital literacy constrain adoption.This study introduces the Intelligent Farmer Assistant and Crop Lifecycle Management Platform, a comprehensive mobile-first solution designed to provide end-to-end agricultural support. The system integrates artificial intelligence (AI), machine learning (ML), and natural language processing (NLP) to assist farmers throughout the crop lifecycle. The platform is implemented using Flutter for cross-platform mobile development, Spring Boot for backend services, and Firebase for real-time data synchronization.The system comprises six key components: (1) a convolutional neural network (CNN)-based crop disease detection module, (2) a multilingual conversational assistant, (3) a long short-term memory (LSTM)-based market price prediction model, (4) a crop lifecycle management system, (5) a location-aware weather advisory module, and (6) a carbon footprint estimation tool. The disease detection model leverages MobileNetV2 with transfer learning, trained on the PlantVillage dataset containing over 87,000 labeled images, achieving 96.5% classification accuracy. The chatbot supports Telugu, Hindi, and English through both text and voice interfaces, with an intent recognition accuracy of 93.7%. The market prediction module utilizes a stacked LSTM architecture trained on AgMarkNet data, achieving strong predictive performance (R² = 0.91, MAPE = 4.2%).