Enhancing Crop Production Leveraging ResNet-152 and Deep Learning Architectures for the Detection of Potato Plant Leaf Diseases in Agriculture

Abstract

This study focuses on the critical issue of identifying and classifying potato plant leaf diseases, a significant concern for agricultural productivity worldwide. Potato crops are susceptible to various diseases, such as early blight and late blight, which can severely affect yield if not accurately identified and treated promptly. The traditional methods for disease identification, primarily manual inspection, are time-consuming and subject to human error, creating a gap in efficient and scalable disease management strategies. To address this gap, we introduced and evaluated several machine learning models, including CNNs, VGG19, ResNet-50, Xception, Inception V3, MobileNet, YOLOv7, and a proposed ResNet-152 model, for their ability to automatically detect and classify potato plant leaf diseases from images. Our method leverages deep learning techniques to enhance the accuracy, precision, recall, and F1 scores of disease identification, offering a significant improvement over existing approaches. The results demonstrate the proposed ResNet-152 model's superior performance, achieving an accuracy of 99.76%, precision of 99.5%, recall of 99.8%, and an Average F1 score of 99.6%. These metrics surpass those of other evaluated models, highlighting the effectiveness of the proposed model in identifying potato plant leaf diseases with high reliability. The study underscores the potential of advanced deep learning architectures to revolutionize agricultural practices by providing a robust tool for early disease detection and management, thereby contributing to increased crop production and reduced losses.