An AI-Driven Smart Farming System for Crop Recommendation and Plant Disease Detection

Abstract

Climate variability, soil degradation, water scarcity, and plant diseases are all contributing to new challenges in agriculture. Conventional approaches often rely on hands-on monitoring and intuition, leading to suboptimal decision-making and lower throughput. However, existing Machine Learning (ML)-based systems for crop recommendation and disease detection face challenges such as limited scope due to the lack of real-time data integration, dependency on a limited dataset, and reduced accuracy under different environmental conditions. To address these challenges, the present study proposes an Artificial Intelligence (AI)-based smart farm system that integrates ML, Deep Learning (DL), and Internet of Things (IoT) technologies. Random Forest (RF) and Support Vector Machine (SVM) are widely employed ML algorithms for crop recommendation based on environmental parameters such as soil nutrients, temperature, humidity, and rainfall. Image-based plant disease detection is conducted through a Convolutional Neural Network (CNN)-based model, allowing for early diagnosis by analyzing the leaves' image. The experimental results reveal that the proposed system yields high performance. In addition, the crop recommendation accuracy ranges between 98% and 99%, with RF performing best on average, while the proposed CNN-based model achieved a disease detection accuracy ranging from 95% to 97%. This integration of real-time data and intelligent decision support drives farm efficiency, cuts pesticide overuse, and maximizes crop yield.