PMS-AI: An Intelligent Plant Monitoring System Integrating Deep Learning and Precision Agriculture Techniques

Abstract

Plant diseases are a major danger to the world’s food security since they result in large agricultural losses every year. Manual inspection techniques used today are time-consuming, labor-intensive, and frequently unreliable. Using deep learning and computer vision, this research presents PMS-AI, an automated plant disease diagnosis system designed to overcome these difficulties. In order to achieve high accuracy and computational economy, the system makes use of MobileNetV2, a lightweight convolutional neural network (CNN). We use extensive data augmentation (rotation, flipping, zooming, and color modifications) and five-fold cross-validation to improve the resilience of the model. Thirty-eight disease classes from 14 plant species were used to train the model, which had an average accuracy of 98.5% with precision and recall rates higher than 98%. Real-time disease diagnosis that works with edge devices (such as cellphones and drones) is one of PMS-AI’s main benefits. lower processing demands in contrast to more conventional CNN models like ResNet and VGG. Scalability for a range of disease kinds and crop species. This technology greatly enhances crop management and early disease intervention by giving farmers a quick, accurate, and affordable diagnostic tool. IoT integration and field deployment in a range of environmental circumstances will be the main topics of future improvements.