Citrus Disease Detection & Classification using Deep Learning and Machine Learning Models

Authors: Palak Singh Parihar, Rajni Mishra

Certificate: View Certificate

Abstract

Citrus crops are vital contributors to the global agricultural economy, but they are susceptible to various diseases that can significantly impact yield and quality. Early detection and accurate classification of these diseases are essential for effective disease management and crop protection. In recent years, deep learning techniques have emerged as powerful tools for image-based disease detection in plants. In this study, we propose a deep learning-based approach for citrus disease detection and classification using state-of-the-art convolutional neural network (CNN) models. Specifically, we explore the efficacy of popular CNN architectures such as VGG16, ResNet, and Inception for automating the detection and classification of citrus diseases from leaf images. We train and evaluate these models on a large dataset of labelled citrus leaf images representing various disease classes. Experimental results demonstrate the effectiveness of the proposed approach in accurately identifying different citrus diseases, achieving high classification accuracies compared to traditional methods. The developed deep learning models offer promising potential for real-time monitoring and early intervention in citrus orchards, thereby contributing to improved disease management practices and sustainable citrus

Introduction

Pests and diseases are the two main features that affect the yield of citrus. There are many types of citrus pests in the wild. Some of them have similar appearances, making it complicated for farmers to identify them exactly in time. In current years, the development of machine learning algorithms has greatly improved the latest technological level of computer vision. These new network structures enable researchers to achieve high exactness in image classification, object detection and semantic segmentation [1]. Then, some studies use machine learning models to identify image-based disease categories. As an important part of the overall agricultural economy, the citrus industry requires citrus plantations to take appropriate disease control measures to minimize losses

Conclusion

A framework based on deep metric learning that can effectively identify citrus ailment from leaf images. The planned architecture includes an embedded module, a cluster prototype module, or a simple neural network classifier for performing disease recognition. The frame also includes a method of generating patches from blade images to further enhance performance. The proposed method involves several steps, including image capture, image segmentation, feature drag, feature selection, and classification. Almost all disease diagnostic techniques have achieved success with disease classification. The evaluation of our process with other deep complex baselines in terms of time competence shows equivalent or superior performance with other baselines. In addition, our framework shows better classification exactness than all other baselines. The most difficult and challenging part is image segmentation of citrus disease, which is further divided into two parts: background part location and extraction and then separation of disease part. In this article, we used pre-trained deep learning models like VGG16, Inception Net, ResNet, NasNet, MobileNet and CNN used for recognition of the citrus disease model.