A Machine Learning Approach To Real Time Crop Recommendation, Plant Disease Identification, And Yield Estimation

This project presents an integrated artificial intelligence system designed to address key challenges in modern agriculture: suboptimal crop selection, plant disease outbreaks, and inaccurate yield forecasting. The system synergistically combines Internet of Things (IoT) sensors, machine learning, and deep learning models into a unified framework to empower farmers with data-driven decision-making. The pipeline is structured into three core modules. First, a Crop Recommendation Engine employs a Random Forest (RF) algorithm, which demonstrated superior performance (99.09% accuracy) over comparative models like k-Nearest Neighbors and Decision Trees, to suggest the most suitable crops based on real-time analysis of soil NPK (Nitrogen, Phosphorus, Potassium) content, pH levels, moisture, temperature, and humidity. Second, a Disease Identification System utilizes a Convolutional Neural Network (CNN) trained on the PlantVillage dataset to accurately detect and classify diseases from leaf images, enabling early intervention. Third, a Yield Prediction Module implements a Decision Tree model to estimate agricultural output using historical and environmental data, including rainfall, temperature, and pesticide usage. This work demonstrates the practical viability of leveraging contemporary AI tools—such as CNNs, Random Forests, and IoT connectivity—to enhance farm productivity, optimize resource utilization, and promote sustainable agricultural practices in an accessible and scalable manner.