AI-Based Crop Recommendation for Farmers and Farm Management System

Abstract

Agricultural decision-making has historically depended on empirical intuition, localized traditional knowledge, and fragmented external advisories. Such manual heuristics frequently fail to account for dynamic soil chemistry, erratic microclimatic fluctuations, and volatile market conditions, leading to misallocated agricultural inputs, suboptimal crop yields, and unmitigated phytopathological losses. This paper presents AgriSense, an integrated software platform engineered to unify machine learning inference pipelines with operational farm management capabilities. The proposed platform couples three dedicated analytical models: an ensemble Random Forest classifier for soil- and climate-informed crop suitability recommendation; an Extreme Gradient Boosting (XGBoost) regression architecture for multi-factorial crop yield forecasting; and a deep residual convolutional neural network (ResNet-50) for automated leaf disease identification, providing classification outputs alongside diagnostic confidence and infection severity estimates. To support agronomic operations, these machine learning services are embedded alongside real-time meteorological ingestion, wholesale commodity pricing intelligence, labor and resource allocation scheduling, a community knowledge forum, an automated conversational assistant, and local supplier marketplace directories. AgriSense is implemented using a decoupled microservice topology comprising a Next.js presentation tier, a Node.js business logic gateway, an asynchronous FastAPI machine learning inference engine, and a PostgreSQL relational persistence store. This study details the structural design, mathematical formulations, algorithmic workflow, and modular integration of AgriSense, establishing a formal empirical evaluation framework for subsequent field verification.