Precision Agriculture: A Generative AI Approach to Leaf Disease Detection and Pesticide Management

Abstract

Recent breakthroughs in agricultural AI have enabled powerful solutions for plant health monitoring that go beyond disease classification to include tailored pesticide recommendations. Central to many modern systems is a Convolutional Neural Network (CNN) trained on high-resolution leaf images, which routinely achieves 95–99 % accuracy across multiple crops and disease types To enhance performance—particularly when labeled data are limited—Generative Adversarial Networks (GANs) are used to augment datasets with realistic synthetic images, leading to notable gains in classification accuracy .Complementing this diagnostic capability, integrated generative modules leverage CNN-derived features and agronomic parameters (e.g., crop type, growth stage, and environmental conditions) to support precision pesticide recommendations . These systems not only suggest the appropriate pesticide type but also calculate optimal dosages, minimizing agrochemical overuse and aligning with sustainable farming practices. Evaluations conducted on crops like soybean and cashew demonstrate impressive outcomes—disease classification rates of 95 – 99 % and effective, context-aware pesticide advice validated by domain experts  Overall, this Generative AI framework integrates high-accuracy disease detection with intelligent, eco-conscious intervention strategies, offering a scalable and robust toolbox for proactive and efficient crop management.