Crop Disease Identification Research Articles

Image based Chili Crop Disease Prediction Using Deep Transfer Learning

Crop diseases have a terrible impact on food protection and can result in considerable reductions in both the supply and quality of agricultural products. Human professional have traditionally been relying on to diagnose crop diseases caused by insects, pests, virus, bacteria, fungal, inadequate nutrition, or adverse environmental conditions. This, however, is costly, time demanding, and in some situations unworkable. Thus, in the area of agricultural information, the automatic identification of crop diseases is significantly required. Many strategies have been presented to solve this challenge, with deep learning becoming as the preferred approach due to its outstanding performance.This research describes a method for detecting chili leaf diseases using a deep convolutional neural network. We compared performances of four architectures: MobileNet, Inception-ResnetV2, EfficientNetB0, and DenseNet. The proposed approach evaluated the findings using measures such as accuracy, loss and time. Our model compares favorably to EfficientNetB0 with an accuracy of 0.995, a loss of 0.023, and time is 5 minutes 45 seconds. EfficientNetB0, a compact deep learning architecture has fine tuned to classify two forms of chili leaf diseases. The method was tested on 2475 photos from the Plant Village dataset.

Crop disease recognition based on the YOLOv5 and transfer learning

Crop disease detection is an important factor in agricultural production. Traditional object detection methods can't effectively screen key features, resulting in weak crop disease control in many countries. In recent years, several convolutional neural networks for object detection have been proposed, which makes it possible to apply computer vision to crop disease identification through deep learning. YOLOv5 is an advanced object detection network, which can extract key features and use human visual attention mechanism as a reference. This paper would evaluate the performance of four pre-trained models of YOLOv5 in object detection of crop diseases. And transfer learning was used to train the corresponding dataset. The experiment results have showed that the F1 values of the four models all reached above 0.93, and the Yolov5x got the best result, which achieved 0.963. Furthermore, the detection accuracy of the four models has reached more than 98%. This shows that the YOLOv5 series network models have great application prospects in the identification of crop diseases. In the near future, the object detection model can be applied to various mobile devices, even unmanned aerial vehicles, which would play a significant role in crop disease prevention.

A review of plant disease detection and classification methods

ABSTRACT The majority of agricultural yields are affected by plant diseases leading to economic losses. Hence, it is necessary to evaluate plant diseases at the very beginning to enhance crop production by making effective control actions. Therefore, automatic plant disease detection models are required to reduce the huge laborious tasks. In recent times, deep learning and computer vision models have been established for the automated detection of plant disease which functions by analyzing the symptoms on plant leaves. Thus, this paper reviews a set of articles that utilizes deep learning and image processing in the field of detecting plant and crop diseases. This survey ensures a comprehensive view of all research works regarding plant disease detection for exploring the effectiveness of the existing models for detecting diseases in diverse crops. This paper aims to analyze various machine learning and deep learning approaches regarding crop disease identification. This review gives an introduction to plant leaf diseases, the conventional methods for detecting plant diseases, the dataset used, the performance metrics utilized for analyzing the existing detection frameworks, and the implementation platforms. Finally, this review provides the limitations in this field of plant leaf detection and the future research trends for further enhancement.

Efficient Disease Identification Method for Crop Leaf using Deep Learning Techniques

Many prime grain-producing nations have implemented steps to limit export of grains as COVID-19 has expanded over the globe; food security has sparked significant worry from a number of stakeholders. One of the most crucial concerns facing all nations is how to increase grain output. However, the diseases occur in crops remain a challenge for countless farmers, therefore it is critical to understand their severity promptly and precisely to guide the them in taking additional measures to lessen the chances of plants being affected furthermore.
 This paper describes a deep learning model for the identification of crop diseases that can achieve high accuracy with low processing power. The model, called the inception v3 network, has been tested on a tomato leaf dataset and has obtained a average identification accuracy of 98.00% and further the ensemble of two inception v3 models with slight diversity achieved an accuracy of 98.11%. The results suggest that this model could be useful in improving food security by helping farmers quickly and accurately identify crop diseases and take appropriate measures to prevent further spread.

Improved MobileNetV2 crop disease identification model for intelligent agriculture.

Using intelligent agriculture is an important way for the industry to achieve high-quality development. To improve the accuracy of the identification of crop diseases under conditions of limited computing resources, such as in mobile and edge computing, we propose an improved lightweight MobileNetV2 crop disease identification model. In this study, MobileNetV2 is used as the backbone network for the application of an improved Bottleneck structure. First, the number of operation channels is reduced using point-by-point convolution, the number of parameters of the model is reduced, and the re-parameterized multilayer perceptron (RepMLP) module is introduced; the latter can capture long-distance dependencies between features and obtain local a priori information to enhance the global perception of the model. Second, the efficient channel-attention mechanism is added to adjust the image-feature channel weights so as to improve the recognition accuracy of the model, and the Hardswish activation function is introduced instead of the ReLU6 activation function to further improve performance. The final experimental results show that the improved MobilNetV2 model achieves 99.53% accuracy in the PlantVillage crop disease dataset, which is 0.3% higher than the original model, and the number of covariates is only 0.9M, which is 59% less than the original model. Also, the inference speed is improved by 8.5% over the original model. The crop disease identification method proposed in this article provides a reference for deployment and application on edge and mobile devices.

Biorthogonal wavelet based entropy feature extraction for identification of maize leaf diseases

Crop disease is considered as a major constraint to both food quality and production. Even in this era of precision agriculture, the lacking of compulsory infrastructure has made rapid identification of crop diseases quite hard in numerous regions around the world. In this paper, we introduced a new method based on biorthogonal wavelet transform (BWT) to identify prime maize leaf diseases. We performed biorthogonal wavelet decomposition and pixel wise morphological operation to segment the maize leaf lesion from input image. For feature extraction, by applying 2-D biorthogonal wavelet transform (BWT) at multiple levels we proposed a novel method to extract colour channel wise wavelet entropy features by investigating discriminatory potential of three different biorthogonal wavelet filters (bior3.3, bior3.5, and bior3.7). The effectiveness of our extracted features were evaluated by employing five different classifiers and obtaining 95.78% overall identification accuracy with 10-fold cross validation. All the materials related our study can be found at: https://github.com/BadhanMazumder/BiorthogonalWavelet_MaizeDiseaseDetection.git.

Application Progress of UAV-LARS in Identification of Crop Diseases and Pests

Disease and pest stress is one of the important threats to crop growth and development, which have a very adverse impact on crop yield and quality every year, even leading to crop failure. Currently, the use of plant protection unmanned aerial vehicles (UAVs) for pesticide spraying is the most effective means of controlling crop diseases and pests. However, the areas where crop pests and diseases occur are often presented in “point-like” or “patchy” forms, and the UAV’s full-coverage spraying method results in a great waste of pesticides. Therefore, there is an urgent need to research a method for identifying the area of diseases and pest stress to achieve precise targeted spraying, in order to reduce the amount of pesticides used and improve their utilization. By analyzing the low-altitude remote sensing images of crop fields taken by UAVs, real-time pesticide spraying prescription maps can be generated to meet the demand for precise targeted spraying. This review focuses on the actual needs of precise targeted spraying by plant protection UAVs. Firstly, the RS monitoring mechanism of crop diseases and pests by UAVs is studied. Secondly, a comprehensive investigation of the literature on UAV Low-altitude Remote Sensing (UAV-LARS) technology for monitoring and identifying crop diseases and pests is conducted, summarizing the research progress in monitoring and identifying crop diseases and pests, especially in wheat, cotton, and rice. Finally, the key issues to be addressed and the future development direction of UAV-LARS monitoring of crop diseases and pests are proposed.

Detection of citrus pests in double backbone network based on single shot multibox detector

To prevent and control agricultural diseases and insect pests, the timely detection and accurate identification of crop diseases and insect pests are significant. Studies have shown that pests on plant surfaces are challenging to detect because of their small size and strong camouflage. Therefore, to better detect pests on citrus leaves, a citrus disease and insect pest detection method based on a double backbone network is proposed. The double backbone network-improved Single Shot MultiBox Detector (SSD) model was used to detect citrus images. The accuracy and recall rate of the neural network target detection were evaluated, and the robustness was verified by analyzing the detection results. The experimental results showed that the trained network’s mean average precision (mAP) on the test dataset was 72.54%. In addition, the model showed high robustness on citrus pest datasets, with mAP reaching 86.01%. The results showed that the method was accurate and efficient compared with other target detection methods and could be applied to detect and control citrus pests and diseases.

Prediction of Disease in Tomato Leaves with use of Machine Learning Technique

India's sizable agricultural market provides the perfect conditions for cultivating a variety of products, including the tomato harvest. Detecting the transmission of diseases from unhealthy to healthy plants poses a severe threat to the agricultural industry because, if caught early enough, they can quickly spread and perhaps infest the entire farm. In terms of profit in good forming, early stage crop disease identification and severity monitoring is quite important. K-means clustering with fuzzy logic is used in the proposed study to evaluate the disease-affected region of the leaf and, as a result, assess the severity of the diseases. In this thesis, illnesses are detected using machine learning models for convolutional neural networks (CNN) and K-nearest neighbours (KNN).

Spectroscopic detection of wheat yellow mosaic virus infection based on invariant shape spectral processing and machine learning

Wheat yellow mosaic disease (WYMD) is a low-temperature soil-borne disease that causes serious yield and economic losses. Reflectance spectroscopy has great potential for detecting crop diseases, but it has not been applied for detection of WYMD. Herein, we collected 2 years of wheat physiological and reflectance spectroscopy data at green-up and jointing stages, and conducted disease detection studies. We evaluated standard normal variate (SNV), multiplicative scatter correction (MSC), and spectral separation of soil and vegetation (3SV) as preprocessing methods for invariant spectral shapes, and 96 spectral indices. The spectral index wavelength (SIW), threshold method, and variance inflation factor (TVIF) were used as feature selection methods combined with machine learning to construct a systematic disease detection scheme. The results showed that the visible region centred at 750 nm was the most sensitive wavelength. Additionally, 900–––1000 nm was the sensitive wavelength in the middle and latter stages, during disease aggravation and fertility development. The preprocessing method was best with 3SV, which was highly consistent with the characteristic regions of the original spectra, followed by SNV and MSC. spectral indices OR-BG1 I, SNV-RPMI, MSC-RPMI and 3SV-CTRI 1 could be applied to detect WYMD, and the best index was 3SV-CTRI 1, with OA and Kappa values of 75.88% and 0.516, respectively. The effectiveness of the SIW feature selection method and coupling with machine learning was superior to TVIF. The best results among different machine learning models were obtained with SVM, followed by RF and KNN. The optimal 3SV-SIW-SVM disease detection mode achieved OA and Kappa values of 94.71% and 0.893, respectively, for field data validation, and 68.06% and 0.381, respectively, for simulated satellite data validation. The findings will facilitate migration of ground-based research to satellite platforms, and provide new ideas for quantitative identification of crop diseases.

An Analysis of Plant Diseases Identification Based on Deep Learning Methods.

Plant disease is an important factor affecting crop yield. With various types and complex conditions, plant diseases cause serious economic losses, as well as modern agriculture constraints. Hence, rapid, accurate, and early identification of crop diseases is of great significance. Recent developments in deep learning, especially convolutional neural network (CNN), have shown impressive performance in plant disease classification. However, most of the existing datasets for plant disease classification are a single background environment rather than a real field environment. In addition, the classification can only obtain the category of a single disease and fail to obtain the location of multiple different diseases, which limits the practical application. Therefore, the object detection method based on CNN can overcome these shortcomings and has broad application prospects. In this study, an annotated apple leaf disease dataset in a real field environment was first constructed to compensate for the lack of existing datasets. Moreover, the Faster R-CNN and YOLOv3 architectures were trained to detect apple leaf diseases in our dataset. Finally, comparative experiments were conducted and a variety of evaluation indicators were analyzed. The experimental results demonstrate that deep learning algorithms represented by YOLOv3 and Faster R-CNN are feasible for plant disease detection and have their own strong points and weaknesses.

VLDNet: An Ultra-Lightweight Crop Disease Identification Network

Existing deep learning methods usually adopt deeper and wider network structures to achieve better performance. However, we found that this rule does not apply well to crop disease identification tasks, which inspired us to rethink the design paradigm of disease identification models. Crop diseases belong to fine-grained features and lack obvious patterns. Deeper and wider network structures will cause information loss of features, which will damage identification efficiency. Based on this, this paper designs a very lightweight disease identification network called VLDNet. The basic module VLDBlock of VLDNet extracts intrinsic features through 1 × 1 convolution, and uses cheap linear operations to supplement redundant features to improve feature extraction efficiency. In inference, reparameterization technology is used to further reduce the model size and improve inference speed. VLDNet achieves state-of-the-art model (SOTA) latency-accuracy trade-offs on self-built and public datasets, such as equivalent performance to Swin-Tiny with a parameter size of 0.097 MB and 0.04 G floating point operations (FLOPs), while reducing parameter size and FLOPs by 297 times and 111 times, respectively. In actual testing, VLDNet can recognize 221 images per second, which is far superior to similar accuracy models. This work is expected to further promote the application of deep learning-based crop disease identification methods in practical production.

Crop Disease Identification by Fusing Multiscale Convolution and Vision Transformer.

With the development of smart agriculture, deep learning is playing an increasingly important role in crop disease recognition. The existing crop disease recognition models are mainly based on convolutional neural networks (CNN). Although traditional CNN models have excellent performance in modeling local relationships, it is difficult to extract global features. This study combines the advantages of CNN in extracting local disease information and vision transformer in obtaining global receptive fields to design a hybrid model called MSCVT. The model incorporates the multiscale self-attention module, which combines multiscale convolution and self-attention mechanisms and enables the fusion of local and global features at both the shallow and deep levels of the model. In addition, the model uses the inverted residual block to replace normal convolution to maintain a low number of parameters. To verify the validity and adaptability of MSCVT in the crop disease dataset, experiments were conducted in the PlantVillage dataset and the Apple Leaf Pathology dataset, and obtained results with recognition accuracies of 99.86% and 97.50%, respectively. In comparison with other CNN models, the proposed model achieved advanced performance in both cases. The experimental results show that MSCVT can obtain high recognition accuracy in crop disease recognition and shows excellent adaptability in multidisease recognition and small-scale disease recognition.

Crop disease identification segmentation algorithm based on Mask‐RCNN

AbstractLeaf diseases include various diseases like abscission, gummosis, Alternaria solani, vesicatoria, and Pucciniales. Farmers from all around the world are constantly concerned about the health of various plants. Our proposed technology identifies various leaf diseases at an early stage, alerting farmers and allowing them to take necessary control measures. The collection contains 25,362 photos of normal leaves, bacterial spot, rust, early blight, and other disease‐infected leaves in Solanum tuberosum, Solanum lycopersicum, Malus, Zea mays, and Piper nigrum. In this study, a transfer learning ResNet 50 algorithm and a mask regional convolutional neural network are used to segment leaf diseases. An improved faster mask regional‐convolutional neural network (IFMR‐CNN) method is proposed to find the affected areas in the plant. The experiment first gathers photos of leaf disease for preprocessing, then utilizes a VGG annotator to create labels for the data sets, which are split into a test set and training set. The proposed IFMR‐CNN approach is capable of localizing and classifying the disease with 96% accuracy. Accuracy is calculated based on the proportion of properly identified leaf samples in a dataset out of all the leaf samples. Accuracy of disease localization in plant leaves can be measured by the percentage of correctly identified regions of interest in the leaf image that contain the disease symptoms.

Performance analysis of AI-based solutions for crop disease identification, detection, and classification

Agriculture is an integral part of human civilization. Apart from providing Food, Agriculture also contributes to the economy. During crop production, crops are susceptible to insects. The identification and treatment of these insects have become a severe issue. The optimal way to reduce this cost is by early detection of the insects and taking appropriate measures to minimize the damage to crop plants. However, the traditional method lacks to examine the diseases and insects' presence in real-time. This study provides deep learning-based solutions for real-time identification and detection of insects in the Soybean crop. Various Transfer-learning models’ performances are examined to explore the proposed solution's feasibility and reliability to find the insect's identification and detection accuracy. The accuracy achieved with the proposed solution is 98.75%, 97%, and 97% using YoloV5, InceptionV3, and CNN, respectively. Among them, the YoloV5 algorithm's performance in the solution is very fast and can run at 53fps, making them fit for real-time detection. Moreover, a dataset of crop insects was collected and labeled by mixing the images collected using different devices. The proposed study helps reduce the producer's workload and is much simpler, and provides better results.

English

Artificial Intelligence (AI) and deep learning have the capacity to reduce losses in crop production, such as low crop yields, food insecurity, and the negative impacts on a country’s economy caused by crop infections. This study aims to find the knowledge and technological gaps associated with the application of AI-based technologies for plant disease detection and pest prediction at an early stage and recommend suitable curative measures. An evidence-based framework known as the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) methodology was used to conduct systematic reviews of the state-of-the-art of AI and deep learning techniques for crop disease identification and pest prediction in developing countries. The results demonstrate that conventional methods for plant disease management face some challenges, such as being costly in terms of labour, having low detection and prediction accuracy, and some are not environmentally friendly. Also, the rapid increase in data-intensive and computational-intensive tasks needed for plant disease classification using traditional machine learning methods poses challenges such as high processing time and storage capacity. Consequently, this paper recommends a deep learning and AI-based strategy to enhance the detection, prediction and prevention of crop diseases. These recommendations will be the starting point for future research. Key words: Plant diseases detection, pest prediction, pesticide recommendation, artificial intelligence, machine learning.

An Effective Pyramid Neural Network Based on Graph-Related Attentions Structure for Fine-Grained Disease and Pest Identification in Intelligent Agriculture

In modern agriculture and environmental protection, effective identification of crop diseases and pests is very important for intelligent management systems and mobile computing application. However, the existing identification mainly relies on machine learning and deep learning networks to carry out coarse-grained classification of large-scale parameters and complex structure fitting, which lacks the ability in identifying fine-grained features and inherent correlation to mine pests. To solve existing problems, a fine-grained pest identification method based on a graph pyramid attention, convolutional neural network (GPA-Net) is proposed to promote agricultural production efficiency. Firstly, the CSP backbone network is constructed to obtain rich feature maps. Then, a cross-stage trilinear attention module is constructed to extract the abundant fine-grained features of discrimination portions of pest objects as much as possible. Moreover, a multilevel pyramid structure is designed to learn multiscale spatial features and graphic relations to enhance the ability to recognize pests and diseases. Finally, comparative experiments executed on the cassava leaf, AI Challenger, and IP102 pest datasets demonstrates that the proposed GPA-Net achieves better performance than existing models, with accuracy up to 99.0%, 97.0%, and 56.9%, respectively, which is more conducive to distinguish crop pests and diseases in applications for practical smart agriculture and environmental protection.

Detection of Diseased Plant Leaf Using Deep Learning

The detection of illness in crops is one of the laborious and crucial duties in agricultural activities. It requires a lot of time and requires specialized labor. Detection of plant disease by use of the automatic technique is beneficial and it requires a huge amount of work, monitoring in the large farm of crops, and at an early stage itself to detect symptoms of diseases affected means where they appear on the plant leaves. A more recent, cutting-edge method for processing images that produce exact outcomes is deep learning. Leaf detection of disease and categorization employ a variety of deep learning and image processing approaches. For disease detection, image processing methods including image pre-processing, segment, feature extraction, etc. are employed along with deep learning methods like CNN, Fast RCNN, Faster RCNN, and Mask RCNN. Furthermore, deep learning technologies offer greater precision than image processing technology. Application areas for the detection of plant leaf diseases include biological research and agricultural institutes, among many others. The economy heavily depends on agricultural output. In this study, an effective method for crop disease identification by the use of machine learning and image processing techniques is proposed. This proposed approach has a 75% accuracy rate for detecting 20 distinct illnesses in 4 popular types of plants. We are using integrated techniques to improve the accuracy above 75% through this model.

Dual-branch collaborative learning network for crop disease identification.

Crop diseases seriously affect the quality, yield, and food security of crops. redBesides, traditional manual monitoring methods can no longer meet intelligent agriculture's efficiency and accuracy requirements. Recently, deep learning methods have been rapidly developed in computer vision. To cope with these issues, we propose a dual-branch collaborative learning network for crop disease identification, called DBCLNet. Concretely, we propose a dual-branch collaborative module using convolutional kernels of different scales to extract global and local features of images, which can effectively utilize both global and local features. Meanwhile, we embed a channel attention mechanism in each branch module to refine the global and local features. Whereafter, we cascade multiple dual-branch collaborative modules to design a feature cascade module, which further learns features at more abstract levels via the multi-layer cascade design strategy. Extensive experiments on the Plant Village dataset demonstrated the best classification performance of our DBCLNet method compared to the state-of-the-art methods for the identification of 38 categories of crop diseases. Besides, the Accuracy, Precision, Recall, and F-score of our DBCLNet for the identification of 38 categories of crop diseases are 99.89%, 99.97%, 99.67%, and 99.79%, respectively. 811.

Automation of Crop Disease Detection through Conventional Machine Learning and Deep Transfer Learning Approaches

With the rapid population growth, increasing agricultural productivity is an extreme requirement to meet demands. Early identification of crop diseases is essential to prevent yield loss. Nevertheless, it is a tedious task to manually monitor leaf diseases, as it demands in-depth knowledge of plant pathogens as well as a lot of work, and excessive processing time. For these purposes, various methods based on image processing, deep learning, and machine learning are developed and examined by researchers for crop leaf disease identification and often have obtained significant results. Motivated by this existing work, we conducted an extensive comparative study between traditional machine learning (SVM, LDA, KNN, CART, RF, and NB) and deep transfer learning (VGG16, VGG19, InceptionV3, ResNet50, and CNN) models in terms of precision, accuracy, f1-score, and recall on a dataset taken from the PlantVillage Dataset composed of diseased and healthy crop leaves for binary classification. Moreover, we applied several activation functions and deep learning optimizers to further enhance these CNN architectures’ performance. The classification accuracy (CA) of leaf diseases that we obtained by experimentation is quite impressive for all models. Our findings reveal that NB gives the least CA at 60.09%, while the InceptionV3 model yields the best CA, reaching an accuracy of 98.01%.