Agricultural crop disease recommendation and leaf disease prediction

Authors: Prof. Nandini Gowda P, Spoorthi S, V. Bindu Shree

Abstract: Agriculture is the backbone of many countries, including India, and provides livelihoods to millions of people facing challenges such as climate change and plant disease outbreaks. Through research, a web application has been developed that provides real-time recommendations for crop selection based on various factors such as soil nutrients, temperature, humidity, pH levels, and rainfall. Recent advances in machine learning and artificial intelligence offer promising solutions to these problems, enabling accurate, data-driven decision-making in agriculture. These technologies have the potential to transform how we predict crop yields and detect plant diseases, thus improving agricultural practices. To achieve this, we trained and examined seven machine learning models, Decision tree, Naive Bayes, SVM, Logistic Regression, Random Forest, XGBoost, and KNN. Among these, Random Forest gives the highest accuracy, making it the best choice for crop forecasting. In addition to crop recommendation, the web application also integrates a Plant Disease Identification system using Convolutional Neural Network (CNN). By analysing leaf images, CNN detects and accurately classifies plant diseases, allowing farmers to intervene early to prevent crop losses. This study aims to empower farmers with accessible technology to make informed decisions, improve crop selection, and effectively cure plant diseases. Combining crop recommendation with disease detection, intelligent crop recommendation systems with plant disease detection contribute to sustainable agriculture, economic stability, and food security in India and beyond.

DOI: https://doi.org/10.5281/zenodo.17984980