A Novel Deep Learning Approach for Enhanced Crop Health Management

Abstract

Agriculture has to be efficient, and improving crop yields is always an important task. But crop diseases are a major threat on farmer's income and food security, with negative effects in human health. However, there have been a limited number of studies that focus on disease detection and classification in cruciferous vegetables due to challenges such as poor monitoring disease and the absence of capable data sets for most areas which require extensive computational power. To address these challenges, our manuscript aims to tackle concerns surrounding the identification and detection of diseases in cruciferous vegetables in rural agriculture. We propose using advanced deep learning techniques, specifically Convolutional Neural Networks (CNNs). In this study, we successfully analyzed the 3 different pretrained models suitable for the accurate classification of 7 different classes of cruciferous vegetable disease. The dataset has over 1542 images with 7 different classes. 5 different cruciferous vegetables are considered namely cauliflower, lettuce, spinach, broccoli and cabbage. CNN architectures such as ResNet-50, VGG-19 and MobileNetV2 are considered after thorough research of the existing works. The dataset is used to train these models, which are then assessed for accuracy, precision, recall, and F1 score. ResNet-50 showed the highest accuracy among the three models, achieving 80%. VGG-19 followed with an accuracy of 77%, while MobileNetV2 fell behind with 72%. For future work, we would be interested in integrating real-time photographs and augmented images into the dataset to improve model performance, as well as real-time data collection resolving challenges. Furthermore, by developing a multi-object deep learning model to recognize cruciferous vegetable diseases from groups of leaves rather than just individual leaves, we could go beyond innovation. Such models have scope for more robust application in the management of disease detection of cruciferous vegetables in agriculture, thereby benefiting farmers as well as consumers.

Topics & Concepts