PotatoCare: A Deep Learning-Powered Framework for Automated Detection and Classification of Potato Disease
Abstract
Potato diseases significantly impact agricultural productivity across the globe, resulting in economic losses and food insecurity. Accurate and timely disease identification is vital for effective management and yield maximization. Here, we suggest a deep learning approach to potato disease classification using state-of-the-art convolutional neural network (CNN) models. We investigated the performance of different pre-trained models, including VGG16, VGG19, Densenet201, MobileNetV2, ResNet50, and EfficientNetB3, to develop an optimum classification system. Among the models, EfficientNetB3 achieved an accuracy of 99.82% with good precision of 98.97%, recall of 99.72%, and F1-score of 99.37%, demonstrating the power and reliability of our model. To enhance model performance and generalization, various image preprocessing techniques, data augmentation, and optimization strategies were applied. The proposed approach has effectively distinguished between multiple classes of potato diseases, providing an automated and scalable framework for the early detection of disease. The high classification accuracy proves the model’s potential for real-world deployment in precision agriculture to assist farmers and plant experts in diagnosing plant diseases more easily and reliably.