Conference Paper

CNN Transfer Learning for Automatic Image-Based Classification of Crop Disease

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

As the latest breakthrough in the field of computer vision, deep convolutional neural network(CNN) is very promising for the classification of crop diseases. However, the common limitation applying the algorithm is reliance on a large amount of training data. In some cases, obtaining and labeling a large dataset might be difficult. We solve this problem both from the network size and the training mechanism. In this paper, using 2430 images from the natural environment, which contain 2 crop species and 8 diseases, 6 kinds of CNN with different depths are trained to investigate appropriate structure. In order to address the over-fitting problem caused by our small-scale dataset, we systemically analyze the performances of training from scratch and using transfer learning. In case of transfer learning, we first train PlantVillage dataset to get a pre-trained model, and then retrain our dataset based on this model to adjust parameters. The CNN with 5 convolutional layers achieves an accuracy of 90.84% by using transfer learning. Experimental results demonstrate that the combination of CNN and transfer learning is effective for crop disease images classification with small-scale dataset.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... It posted an accuracy level of 90.4% when assessed using the hold-out test set. In another article, Wang et al. [22] combined transfer learning and CNN to devise a method to classify the images of crop diseases. The authors used a CNN with five convolutional layers and achieved a 90.84% accuracy. ...
... All the residual units are made up of pooling, convolutional, and layers. While ResNet exhibits some similarities with VGG net [22], it runs eight times deeper compared to VGG [28]. The ResNet 18 is the ideal option based on its performance and depth. ...
Article
Full-text available
Plant diseases are a major cause of destruction and death of most plants and especially trees. However, with the help of early detection, this issue can be solved and treated appropriately. A timely and accurate diagnosis is critical in maintaining the quality of crops. Recent innovations in the field of deep learning (DL), especially in convolutional neural networks (CNNs) have achieved great breakthroughs across different applications such as the classification of plant diseases. This study aims to evaluate scratch and pre-trained CNNs in the classification of tomato plant diseases by comparing some of the state-of-the-art architectures including densely connected convolutional network (Densenet) 120, residual network (ResNet) 101, ResNet 50, ReseNet 30, ResNet 18, squeezenet and Vgg.net. The comparison was then evaluated using a multiclass statistical analysis based on the F-Score, specificity, sensitivity, precision, and accuracy. The dataset used for the experiments was drawn from 9 classes of tomato diseases and a healthy class from PlantVillage. The findings show that the pretrained Densenet-120 performed excellently with 99.68% precision, 99.84% F-1 score, and 99.81% accuracy, which is higher compared to its non-trained based model showing the effectiveness of using a combination of a CNN model with fine-tuning adjustment in classifying crop diseases.
... Researchers in the past have extensively utilized transfer learning techniques in their studies due to their ability to train models quickly compared to training from scratch [10]. Moreover, since transfer learning models are pre-trained on large datasets such as ImageNet, they exhibit efficient feature identification processes, leading to improved model accuracy. ...
Conference Paper
Full-text available
Plant diseases are a common and significant problem for farmers worldwide, leading to reduced productivity and eco-nomic challenges for both farmers and countries. Deep learning methods offer an efficient way to classify plant diseases at an earlier stage, enhancing the quality and quantity of agricultural products. Despite the existence of traditional and computer vision classification approaches, they frequently encounter challenges like time-consuming processes, imbalanced data, and restricted field access. This research evaluates several widely used state-of-the-art deep networks on three datasets: PlantVillage, Taiwan dataset, and Citrus Fruits and Leaves Dataset, covering diseases in apple, tomato, and citrus leaves. The evaluation results demonstrate the effective recognition of disease images by deep networks. Notably, the comparison reveals the superiority of specific networks for each dataset: DenseNet201 for PlantVillage - tomato, MobileNetV3 Large for Taiwan dataset - tomato, MobileNetV2 for PlantVillage - apple, and ResNet101 for Citrus Fruits and Leaves Dataset.
... The exploration of network depth emerged as a pivotal factor in classification performance. Wang et al. [10] shed light on this aspect, revealing the interplay between network depth and classification accuracy. Their findings indicated that even with transfer learning, a shallow convolutional architecture could yield commendable classification performance. ...
Article
Full-text available
Leaf disease detection is a crucial task in modern agriculture, aiding in early diagnosis and prevention of crop infections. In this research paper, authors present a comprehensive study comparing nine widely used pre-trained models, namely DenseNet201, EfficientNetB3, EfficientNetB4, InceptionResNetV2, MobileNetV2, ResNet50, ResNet152, VGG16, and Xception, with our newly developed custom CNN (Convolutional Neural Network) for leaf disease detection. The objective is to determine if our custom CNN can match the performance of these established pre-trained models while maintaining superior efficiency. The authors trained and fine-tuned each pre-trained model and our custom CNN on a large dataset of labeled leaf images, covering various diseases and healthy states. Subsequently, the authors evaluated the models using standard metrics, including accuracy, precision, recall, and F1-score, to gauge their overall performance. Additionally, the authors analyzed computational efficiency regarding training time and memory consumption. Surprisingly, our results indicate that the custom CNN performs comparable to the pre-trained models, despite their sophisticated architectures and extensive pre-training on massive datasets. Moreover, our custom CNN demonstrates superior efficiency, outperforming the pre-trained models regarding training speed and memory requirements. These findings highlight the potential of custom CNN architectures for leaf disease detection tasks, offering a compelling alternative to the commonly used pre-trained models. The efficiency gains achieved by our custom CNN can be beneficial in resource-constrained environments, enabling faster inference and deployment of leaf disease detection systems. Overall, our research contributes to the advancement of agricultural technology by presenting a robust and efficient solution for the early detection of leaf diseases, thereby aiding in crop protection and yield enhancement.
... Saliency map method of Brahimi et al [15] to localize infected regions is supposed to improve accuracy for classification. Wang et al [16] studied the effect of the depth of neural network on the accuracy for classification. The biological name of tomatoes is Solanum lycopersicum and this vegetable grows on drained soil [17]. ...
... In addition, some studies developed TL models to classify diseases or insects in various field crops. Wang et al. (2018a) pre-trained CNN-based models with the PlantVillage dataset and fine-tuned them with in-field images to adapt the model for the classification of crop disease in the target area. Paymode and Malode (2022) fine-tuned the VGG model for disease detection of both tomatoes and grapes, which achieved a high accuracy of 98.40% of grapes and 95.71% of tomatoes. ...
Article
Machine learning (ML) has proven to be a powerful tool for utilizing the rapidly increasing amounts of remote sensing data for environmental monitoring. Yet ML models often require a substantial amount of ground truth labels for training, and models trained using labeled data from one domain often demonstrate poor performance when directly applied to other domains. Transfer learning (TL) has emerged as a promising strategy to address domain shift and alleviate the need for labeled data. Here we provide the first systematic review of TL studies in environmental remote sensing. We start by defining the different forms of domain shift and then describe five commonly used TL techniques. We then present the results of a systematic search for peer-reviewed articles published between 2017 and 2022, which identified 1676 papers. Applications of TL in remote sensing have increased rapidly, with nearly 10 times more publications in 2022 than in 2017. Across seven categories of applications (land cover mapping, vegetation monitoring, soil property estimation, crop yield prediction, biodiversity monitoring, water resources management, and natural disaster management) we identify several recent successes of TL as well as some remaining research gaps. Finally, we highlight the need to organize benchmark datasets explicitly for TL in remote sensing for model evaluation. We also discuss potential research directions for TL studies in environmental remote sensing, such as realizing scale transfer, improving model interpretability, and leveraging foundation models for remote sensing tasks.
... Pretrained models are more useful and have higher performance compared to the models trained from scratch. Some researchers (Wang et al. 2018; Mukti and Biswas 2019; Ghosal and Sarkar 2020) employed transfer learning on preexisting models trained on different data to improve the classification accuracy of plant diseases. In the literature, there are various pre-trained CNN models such as GoogleNet (Szegedy et al. 2015), ResNet (He et al. 2016), and DenseNet (Huang et al. 2017). ...
Article
Fusarium wilt of chickpea causes significant losses in chickpea production worldwide due to its severely damaging characteristic. On the other hand, it can be controlled by growing resistant varieties, using fungicides, or by on-site applications. It is very important to detect the disease in the early stages before the disease is fully transmitted to the product in order to avoid the disruption of the production of chickpeas. This study investigated the applicability of pre-trained models based on deep learning, which could help determine the type of infection at an early stage of the Fusarium wilt of chickpea, with a proposed new dataset. The results of the study showed that pre-trained Convolutional Neural Network models could be used for classifying the disease. Models can classify the images of chickpea leaves used as inputs as "Highly Resistant, Resistant, Moderately Resistant/ Tolerant, Susceptible, and Highly Susceptible", according to the severity of the disease. Convolutional Neural Networks are among the state-of-the-art deep learning approaches and one of the approaches that are inspired by the human brain and can automatically learn the distinguishing features from the dataset. Therefore, they can perform close to expert performance in different tasks and applications. The study used a novel dataset containing images of chickpea plants infected by the pathogen Fusarium oxysporum f. sp. ciceris to train, validate, and test the pre-trained deep-learning models. According to the results of the study, DenseNet-201, with an average test accuracy of 90%, outperformed the other models. Furthermore, the confusion matrix of DenseNet-201 shows that the other metrics (precision, recall, and F1-Score) were consistent with the average test accuracy. The obtained accuracy value and other performance indicators indicate that pre-trained Convolutional Neural Network models can help determine the severity of Fusarium wilt in chickpea.
... This type of visualization improves classification accuracy. Wang et al [20] identified the impact of depth of network on classification accuracy. Even with transfer learning, high classification accuracy is achieved with low number of convolutional layers. ...
Chapter
Full-text available
Accurate diagnosis plays an important role in preventing crop quality and high growth of farming. Recently, deep learning technique especially Convolutional Neural Networks (CNN), have achieved excellent results in many applications such as plant disease detection and classification. The state-of-the-art architectures are AlexNet, GoogleNet, Inception V3, Residual Network (ResNet) 18, and ResNet 50. All are used to compare and evaluate. The dataset used for the experiment contains 9 different classes of tomato disease and 1 healthy class of Plant Village open data set by Kaggle. Many works are done in the area to find the healthy and unhealthy plants using CNN. Models were evaluated by multiclass statistical analysis based on accuracy, precision, sensitivity, specificity, F-score, area under the ROC curve and receiver operating characteristic (ROC). The results represent the significant values obtained by the Google-Net method, with an AUC of more accuracy and sensitivity With this high success rate, we can conclude that the GoogleNet model has become a useful tool to help farmers to identify tomatoes and protect them from the aforementioned diseases
... The idea of plant disease classification based on transfer learning is to transfer knowledge from a source domain to a target domain by relaxing the assumption. Wang et al. [10] pretrained the convolutional neural network model in PlantVillage and adjusted the neural network parameters on their own dataset. The classification accuracy of crop disease images in a small dataset reached 90.84%. ...
Article
Full-text available
The effective and accurate classification of plant diseases is an important task for agricultural production. Therefore, some studies have utilized convolutional neural networks to identify categories of plant diseases, which can effectively reduce the reliance on crop experts. To further improve the accuracy of plant disease classification methods based on convolutional neural network, this paper proposes an attention model designed for plant disease classification tasks. The proposed attention model contains bit-plane attention and a correlation spatial attention. The bit-plane attention localizes disease areas by exploiting bit-plane information. The correlation spatial attention enhances the weight of important areas in the feature map by establishing the correlation between different areas. The accuracy of the proposed attention model inserted into ResNet101 on the AI Challenger 2018 and PlantVillage datasets is 87.11% and 99.82%, respectively. The performance is better than that of other methods studied on the public plant disease classification dataset. Experiments show that the proposed attention model outperforms the widely used universal attention models SE, CBAM, CA, ECA, BAM and GC. In addition, ablation experiments are conducted to verify the influences of different variants of the proposed attention model on the results.
... A high classification accuracy can be achieved with a low number of convolution layers even with transfer learning [17]. For learning the parameters from the pictures of the injury, a variable momentum rule to CNN yields converging results with comparatively high accuracy [15]. ...
Conference Paper
Each cuisine required tomato in their kitchen for various food items and this makes tomato most popular crop worldwide and India is in second rank in terms production of tomato. Now a days, production of tomato goes down because of various diseases and to treat these diseases farmer needs to have extensive prior knowledge about the pathogen and along with various factor which promote the disease in the tomato. Due to lack of knowledge, the disease spreads rapidly and destroys all crops. To fill this gap, deep learning (DL) has been playing an important role, and there is much research on DL, how it can be used in medical industry and the agriculture industry for the identification of disease using images. There is a limitation for DL model that it does not work well with small dataset and huge amount of samples are required to train the model. Moreover, the data are not shared openly for security or for any other reason. Therefore, to overcome this challenge a Federated Learning (FL) based approach has been presented in the article. In FL, a deep learning model is shared with organizations which having the data and train the model. After training, the model information is shared with a centralized server which designs a generalized model. After getting the generalized model, it is shared with all other sites. The process is repeated until a generalized model is not designed and well-suited with all the sites. In our study, we tested our model on a tomato leaf disease data set using FL methodology with 10 clients and achieved the best precision with 88. 01%.
... Fang et al. [9] proposed an instance-based transfer learning method to solve the problem of insufficient training samples of agricultural disease images. Wang et al. [10] pretrained on the PlantVillage dataset using CNN and fine-tuned their plant disease dataset. The experimental results show that combining CNN with transfer learning can improve the classification accuracy of small datasets. ...
Article
Full-text available
Plant diseases are a critical threat to the agricultural sector. Therefore, accurate plant disease classification is important. In recent years, some researchers have used synthetic images of GAN to enhance plant disease recognition accuracy. In this paper, we propose a generative adversarial classified network (GACN) to further improve plant disease recognition accuracy. The GACN comprises a generator, discriminator, and classifier. The proposed model can not only enhance convolutional neural network performance by generating synthetic images to balance plant disease datasets but the GACN classifier can also be directly applied to plant disease recognition tasks. Experimental results on the PlantVillage and AI Challenger 2018 datasets show that the contribution of the proposed method to improve the discriminability of the convolution neural network is greater than that of the label-conditional methods of CGAN, ACGAN, BAGAN, and MFC-GAN. The accuracy of the trained classifier for plant disease recognition is also better than that of the plant disease recognition models studied on public plant disease datasets. In addition, we conducted several experiments to observe the effects of different numbers and resolutions of synthetic images on the discriminability of convolutional neural network.
... VGG-16, a transfer learning model, performed best on the hold-out test set with 90.4 percent accuracy. Wang et al.[28] developed crop disease picture categorization using CNN with transfer learning. CNN with transfer learning of additional convolutional layers improved classification rate to 90.84%. ...
Article
Tomato leaf disease detection has been a challenge for farmers due to the need for quick and accurate diagnosis to prevent plant contamination. Traditional approaches such as visual observation and laboratory testing have limitations such as subjectivity and reliance on human expertise. Deep Learning (DL) approaches, particularly Convolutional Neural Networks (CNN), have emerged as a solution for disease identification and classification in plants. This research proposes a new variant of an inception-based sequential network that combines the advantages of deep residual and dense networks to reduce dimensionality and enhance accuracy, information flow, and gradient. Modified input image characteristics and hyper-parameters were used to classify existing networks on the tomato dataset, with experimental findings showing that the proposed model can achieve an average identification accuracy of up to 96 percent on the tomato test dataset. The model's performance surpasses most existing leaf identification models and requires fewer computation.
... With simpler architecture, CNN precision is better by 8% compared to CLSTM and 5% compared to RCNN. Wang et al. [31] reduced overfitting by reducing the network complexity. The complete result of the comparison can be seen in Table 6. ...
Article
Full-text available
Background: Twitter is one of the most used social media, with 310 million active users monthly and 500 million tweets per day. Twitter is not only used to talk about trending topics but also to share information about accidents, fires, traffic jams, etc. People often find these updates useful to minimize the impact. Objective: The current study compares the effectiveness of three deep learning methods (CNN, RCNN, CLSTM) combined with neuroNER in classifying multi-label incidents. Methods: NeuroNER is paired with different deep learning classification methods (CNN, RCNN, CLSTM). Results: CNN paired with NeuroNER yield the best results for multi-label classification compared to CLSTM and RCNN. Conclusion: CNN was proven to be more effective with an average precision value of 88.54% for multi-label incidents classification. This is because the data we used for the classification resulted from NER, which was in the form of entity labels. CNN immediately distinguishes important information, namely the NER labels. CLSTM generates the worst result because it is more suitable for sequential data. Future research will benefit from changing the classification parameters and test scenarios on a different number of labels with more diverse data. Keywords: CLSTM, CNN, Incident Classification, Multi-label Classification, RCNN
Article
Diagnosis is given top priority in terms of farm resource allocation, because it directly affects the GDP of the country. Crop analysis at an early stage is important for verifying the efficient crop output. Computer vision has a number of intriguing and demanding concerns, including disease detection. After China, India is the world’s second-largest creator of wheat. However, there exist algorithms that can accurately identify the most prevalent illnesses of wheat leaves. To help farmers keep track on a large area of wheat plantation, leaf image and data processing techniques have recently been deployed extensively and in pricey systems. In this study, a hybrid pre-processing practice is used to remove undesired distortions while simultaneously enhancing the images. Fuzzy C-Means (FCM) is used to segment the affected areas from the pre-processed images. The data is then incorporated into a disease classification model using a Convolutional Neural Network (CNN). It was tested using Kaggle data and several metrics to see how efficient the suggested approach was. This study demonstrates that the traditional Long-Short Term Memory (LSTM) technique achieved 91.94% accuracy on the input images, but the hybrid pre-processing model with CNN achieved 95.06 percent accuracy.
Chapter
Full-text available
The health of the plant and the safety of the food are linked closely. Plant diseases become more complex to the farmers if they are not observed in the early stages. This will become more dangerous to the crop yielding and may reduce the production also. Several types of diseases are identified by the researchers that can cause huge losses to the farmers. Machine learning (ML) algorithms are most widely used to detect the accurate patterns of plant diseases but it is very complex to detect the plant diseases accurately. Deep learning (DL) is most widely used to process complex and large datasets efficiently. In this paper, an integrated learning approach (ILA) is introduced to detect plant diseases on leaves. ILA is the approach integrated with noise filters for removing noise from the leaf images and detecting the depth of the infected region from the leaf image. Advanced training is used to train on strong features of plant diseases. This will help us to increase the accurate detection of plant diseases. The experiments are conducted on two publically available datasets as Kaggle plant diseases dataset and the PlantVillage dataset is used and which consists of 87,848 leaf images containing healthy and disease infected plants. The performance is measured by using the sensitivity, specificity, accuracy, and detection rate.
Article
Agriculture is the most important sector among all, as it supplies food to people. As the food demand is increasing in our daily life and the population is increasing there is need of increasingin the food production also. The main issue nowadays is that diseases that are causing the food production to reduce. So our main aim is identification of the disease of the plants in early stage so that we can do the early diagnosis and protect crops from the harmful diseases. By doing this we can avoid the damage of crops and also can reduce the loses that farmers suffer from damage of crops. Here we are using image processing technique to identify the disease of the crops so that we can cure the diseases of the crops in early stage. Here we are using Deep learning algorithms like Convolutional neural networks (CNN) for our project that is leaf disease detection. CNN is the best algorithm for image processing as it greatly increases the performance of our model. Nowadays deep learning is the current trend as it solves the most complicated problems easily and can get the highest accuracy.
Article
Plant disease diagnosis is very important for agriculture because of its importance in increasing crop production. Nowadays the advances in image processing gives a new way. One of the newest way to solve this issue via visual plant disease analysis. In this paper, we discuss the problem of plant disease recognition. Here we tackle plant disease recognition via reweighting both visual regions and loss to emphasize diseased parts. We first compute the weights of all the divided patches from each image based on the cluster distribution of these patches to indicate the discriminative level of each patch. Then we allocate the weight to each loss for each patch-label pair during weakly supervised training to enable discriminative disease part learning. We finally extract patch features from the network trained with loss reweighting, and utilize the LSTM network to encode the weighed patch feature sequence into a comprehensive feature representation. Extensive evaluations on this dataset and another public dataset demonstrate the advantage of the proposed method. We expect this research will further the agenda of plant disease recognition in the community of image processing
Article
Plant disease identification is vital for agriculture because it is essential for enhancing crop yields. Visual plant disease analysis is a novel technique to handle this problem because to recent developments in image processing. There are, however, few works in this field, let alone comprehensive studies. We investigate the challenge of visual plant disease detection for plant disease diagnosis in this research. Plant disease photos, in comparison to other types of photographs, tend to have randomly dispersed lesions, varied symptoms, and complex backgrounds, making discriminative information difficult to capture. We created a new large-scale plant disease dataset containing 271 plant disease categories and 220,592 photos to aid plant disease recognition studies. We approach plant disease recognition using this dataset to emphasise sick portions by reweighting both visual regions and loss to determine the discriminative level of each patch, we first compute the weights of all the divided patches from each image based on the cluster distribution of these patches. Then, during weakly-supervised training, we assign weight to each loss for each patch-label pair to enable discriminative disease component learning. Finally, we use the LSTM network to encode the weighed patch feature sequence into a comprehensive feature representation, extracting patch features from the network trained with loss reweighting. Extensive tests on this and another publicly available datasets show that the proposed strategy is superior. We anticipate that our study will advance the plant disease recognition agenda in the image processing community.
Chapter
Agriculture being the important aspect of India’s economy and also a major source of crop production. A huge number of India’s population depends on yield production for their livelihood. We are living in an era where problem regarding agriculture is a major issue now a days. The major problem in the crop growth is we have to take care of the health of the plants and the crops. Agriculture is one sector that has a significant influence on human lives and economic condition. Agriculture goods are lost due to poor management. Plant leaves which are the most vulnerable are the first to develop illness signs. From the beginning of their life cycle through the moment they are ready to be harvested, crops must be checked for illnesses. The devices used to predict illness have shown to be faster, cheaper, and more accurate than farmers’ conventional approach of manual observation. Disease signs can be noticed on the leaves, stems, and fruits in the majority of instances. Due to the various factors the productivity of the crop is affected. The factors include change in climatic conditions, pest attacks or various diseases a plant suffers from. An automatic detection is designed to detect disease symptoms at an early or growing stage. The paper exposes a methodology for detecting the diseases in leaves with deep learning and image processing techniques.
Article
Objectives The plant tomato (Solanum Lycopersicum) is vastly infected by various diseases. Exact diagnosis on time contributes a significant job to the good production of tomato crops. The key objective of this article is to recognize the infection in tomato leaves with better accuracy and in less time. Methods Nowadays deep convolutional neural networks have attained surprising outcomes in several applications, together with the categorization of tomato leaves infected with several diseases. Our work is based on deep CNN with different residual networks. Finally; we have performed tomato leaves disease classification by using pre-trained deep CNN with the residual network using MATLAB available on the cloud. Results We have used a dataset of tomato leaves for the experiments which contain six different types of diseases with one healthy tomato leaf class. We have collected 6,594 tomato leaves dataset from Plant Village and we did not collect actual tomato leaves for testing. The outcome obtained by ResNet-50 shows a significant result with 96.35% accuracy for 50% training and 50% testing data and if we focus on time consumption for the outcome then ResNet-18 consumes 12.46 min for 70% training and 30% testing. Conclusions After observation of several outcomes, we have concluded that ResNet-50 shows a better accuracy for 50% training and 50% testing of data and ResNet-18 shows better efficiency for 70% training and 30% testing of data for the same dataset on the cloud.
Article
Full-text available
Objectives COVID-19 is frightening the health of billions of persons and speedily scattering worldwide. Medical studies have revealed that the majority of COVID-19 patients. X-ray of COVID-19 is extensively used because of their noticeably lower price than CT. This research article aims to spot the COVID-19 virus in the X-ray of the chest in less time and with better accuracy. Methods We have used the inception-v3 available on the cloud platform transfer learning model to classify COVID-19 infection. The online Inception v3 model can be reliable and efficient for COVID-19 disease recognition. In this experiment, we collected images of COVID-19-infected patients, then applied the online inception-v3 model to automatically extract features, and used a softmax classifier to classify the COVID-19 images. Finally, the experiment shows inception v3 is significant for COVID-19 image classification. Results Our results demonstrate that our proposed inception v3 model available on the cloud platform can detect 99.41% of COVID-19 cases between COVID-19 and Lung Mask diseases in 44 min only. We have also taken images of the normal chest for better outcomes. To estimate the computation power of the model, we collected 6018 COVID-19, Lung Masks, & Normal Chest images for experimentation. Our projected model offered a trustworthy COVID-19 classification by using chest X-rays. Conclusions In this research paper, the inception v3 model available on the cloud platform is used to categorize COVID-19 infection by X-ray images. The Inception v3 model available on the cloud platform is helpful to clinical experts to examine the enormous quantity of human chest X-ray images. Scientific and clinical experiments will be the subsequent objective of this paper.
Preprint
Full-text available
Crop diseases are a major threat to food security and their rapid identification is important to prevent yield loss. Swift identification of these diseases are difficult due to the lack of necessary infrastructure. Recent advances in computer vision and increasing penetration of smartphones have paved the way for smartphone-assisted disease identification. Most of the plant diseases leave particular artifacts on the foliar structure of the plant. This study was conducted in 2020 at Department of Computer Science and Engineering, University of Engineering and Technology, Lahore, Pakistan to check leaf-based plant disease identification. This study provided a deep neural network-based solution to foliar disease identification and incorporated image quality assessment to select the image of the required quality to perform identification and named it Agricultural Pathologist (Agro Path). The captured image by a novice photographer may contain noise, lack of structure, and blur which result in a failed or inaccurate diagnosis. Moreover, AgroPath model had 99.42% accuracy for foliar disease identification. The proposed addition can be especially useful for application of foliar disease identification in the field of agriculture.
Chapter
Cotton is the most essential crop and plays an important role in the agricultural economy of the country. Cotton crop is prone to many diseases because of changes in the climatic conditions, insects such as pink bollworm, and many other factors. These diseases decrease crop productivity, and at present farmers, are diagnosing the diseases with their own experience. But these kinds of naked-eye observations do not give accurate results on large plantation areas. Therefore, it is necessary to develop an automatic, accurate, and economic system for detecting leaf diseases. The aim of this work is to detect the infected cotton leaf using Convolutional Neural Network (ConvNet/CNN) which is a deep learning technique. Nearly 519 healthy leaves and 387 diseased leaves are collected from reliable sources and studied. This work focusses on the performance evaluation and comparison of the powerful CNN architectures: AlexNet, InceptionV3, and Residual Network (ResNet) 50, VGG 16, NASNet and Xception in detecting the diseased cotton leaf. Out of these six models, ResNet50 and VGG 16 has shown significant performance with 97.56% of accuracy.KeywordsCotton leaf diseaseDeep learningTransfer learningConvolutional neural network
Article
With the development of modern agricultural facilities, crop diseases recognition, nutritional status and morphology achieved rapid growth. To avoid yield loss caused by the delay of disease detection, digital images that contain information with respect to crop growth, disease type and nutrition deficiency have been studied by some researchers. However, traditional image processing methods fail to extract typical disease features of crop images with ambiguous disease information. In this paper, a crop disease image recognition technique based on the salient region and hierarchical indexing was proposed. Improved Harris algorithm and maximum radius were used to calculate the widest salient region. In order to eliminate the effect of different salience distribution ranges between different features, a group of images in the cucumber disease image library were normalized. Experiment results indicate that the time complexity of each algorithm will go up as the size of the dataset increase. Especially when testing large datasets, nonhierarchical and nonclustering, hierarchical and nonclustering and hierarchical based on points all tend to raise the algorithm’s time complexity. Plant Village dataset and AI Challenger 2018 dataset were utilized to compare the recognition performances among the models based on machine learning, neural network, deep learning and our methods. The experiment results show that the method proposed in this paper is capable of recognizing local similar images effectively rather than global similar images, therefore, it has better recognition performance than the model learning methods in the early detection stage of crop disease.
Article
Full-text available
Identification of plant disease is usually done through visual inspection or during laboratory examination which causes delays resulting in yield loss by the time identification is complete. On the other hand, complex deep learning models perform the task with reasonable performance but due to their large size and high computational requirements, they are not suited to mobile and handheld devices. Our proposed approach contributes automated identification of plant diseases which follows a sequence of steps involving pre-processing, segmentation of diseased leaf area, calculation of features based on the Gray-Level Co-occurrence Matrix (GLCM), feature selection and classification. In this study, six color features and twenty-two texture features have been calculated. Support vector machines is used to perform one-vs-one classification of plant disease. The proposed model of disease identification provides an accuracy of 98.79% with a standard deviation of 0.57 on tenfold cross-validation. The accuracy on a self-collected dataset is 82.47% for disease identification and 91.40% for healthy and diseased classification. The reported performance measures are better or comparable to the existing approaches and highest among the feature-based methods, presenting it as the most suitable method to automated leaf-based plant disease identification. This prototype system can be extended by adding more disease categories or targeting specific crop or disease categories.
Article
Full-text available
Crop diseases are a major threat to food security and their rapid identification is important to prevent yield loss. Swift identification of these diseases are difficult due to the lack of necessary infrastructure. Recent advances in computer vision and increasing penetration of smartphones have paved the way for smartphone-assisted disease identification. Most of the plant diseases leave particular artifacts on the foliar structure of the plant. This study was conducted in 2020 at the Department of Computer Science and Engineering, University of Engineering and Technology, Lahore, Pakistan to check leaf-based plant disease identification. This study provided a deep neural network-based solution to foliar disease identification and incorporated image quality assessment to select the image of the required quality to perform identification and named it Agricultural Pathologist (Agro Path). The captured image by a novice photographer may contain noise, lack of structure, and blur which result in a failed or inaccurate diagnosis. Moreover, AgroPath model had 99.42% accuracy for foliar disease identification. The proposed addition can be especially useful for application of foliar disease identification in the field of agriculture.
Article
Plant diseases are considered one of the main factors influencing food production and minimize losses in production, and it is essential that crop diseases have fast detection and recognition. The recent expansion of deep learning methods has found its application in plant disease detection, offering a robust tool with highly accurate results. In this context, this work presents a systematic review of the literature that aims to identify the state of the art of the use of convolutional neural networks(CNN) in the process of identification and classification of plant diseases, delimiting trends, and indicating gaps. In this sense, we review 121 papers selected in the last ten years with different approaches to treat aspects related to disease detection, characteristics of the dataset, the crops and pathogens investigated. From the results of the systematic review, it is possible to understand the innovative trends regarding the use of CNN’s in the identification of plant diseases and to identify the gaps that need the attention of the research community.
Chapter
Plant diseases are causing serious damage to the food product quality and quantity. An automatic identification system to detect and classify the plant diseases is the need of the hour, to manage the food production, and to ease the issue of food security. The paper illustrates the detection and classification of image-based plant diseases with the application of different machine learning techniques using transfer learning. The study is conducted on infected and healthy potato leaves from the PlantVillage dataset. The paper demonstrates an empirical comparison of different deep learning transfer models. The transfer learning is performed on different models that include VGG 16, Inception V4, ResNet 50, and DenseNet with 121 layers. VGG16 attains a testing accuracy of 99% and validation accuracy of 95%, which is very high as compared to other architectures. The training of the model is performed on python with Keras. The process of transfer learning requires a pre-trained network on a large dataset.
Chapter
Full-text available
With the advent of the World Wide Web, big data has gained immense popularity among the research community. This data originates from various sources such as e-commerce sites, stock exchange data, social networking sites, and data stored at weather stations through satellite among others. The data generated by these sources is huge and is referred to as big data due to its sheer volume, variety, and velocity. Most of the data generated is unstructured and heterogeneous. The traditional data analytic tools, relational database management systems, and data warehousing systems are insufficient to meet the increasing computing demand arising out of big data. Processing this huge data is a challenging task that involves high storage devices, complex computing, and powerful processing. This paper discusses several challenges posed by big data in the current scenario and studies different approaches to address these challenges.
Preprint
Full-text available
Identification of plant disease is usually done through visual inspection or during laboratory examination which causes delays resulting in yield loss by the time identification is complete. On the other hand, complex deep learning models perform the task with reasonable performance but due to their large size and high computational requirements, they are not suited to mobile and handheld devices. Our proposed approach contributes automated identification of plant diseases which follows a sequence of steps involving pre-processing, segmentation of diseased leaf area, calculation of features based on the Gray-Level Co-occurrence Matrix (GLCM), feature selection and classification. In this study, six color features and twenty-two texture features have been calculated. Support vector machines is used to perform one-vs-one classification of plant disease. The proposed model of disease identification provides an accuracy of 98.79% with a standard deviation of 0.57 on 10-fold cross-validation. The accuracy on a self-collected dataset is 82.47% for disease identification and 91.40% for healthy and diseased classification. The reported performance measures are better or comparable to the existing approaches and highest among the feature-based methods, presenting it as the most suitable method to automated leaf-based plant disease identification. This prototype system can be extended by adding more disease categories or targeting specific crop or disease categories.
Article
Rapid and accurate identification of wheat leaf diseases and their severity is benefit for the precise prevention and control of wheat leaf diseases. Taking powdery mildew and stripe rust as research objects, this study proposes an algorithm for identification of wheat leaf diseases and their severity based on Elliptical-Maximum Margin Criterion (E-MMC) metric learning. Compared with other metrics, elliptic metric combined with MMC can find the non-linear transformation that reflects the spatial structure or semantic information of the wheat leaf disease image, which can enlarge the distance between different classes and better complete the identification task. In the proposed algorithm, Otsu method is used to segment the disease spots according to the characteristics of disease distribution in wheat leaf images. Moreover, the best combination of color and texture features in the wheat disease spot image is determined to construct training set. By using the maximum margin criterion and gradient rise method, the optimal elliptic metric matrix is obtained, thereby transforming the sample feature space and reducing the correlation between features. Then, the wheat powdery mildew, stripe rust, and their severity are identified. The experimental results show that the proposed algorithm is superior to the traditional support vector machines and other algorithms. The highest identification accuracy obtained by the proposed algorithm is 94.16 %. These findings can provide valuable help for the intelligent identification and classification of wheat leaf diseases.
Article
This article has been withdrawn: please see Elsevier Policy on Article Withdrawal (https://www.elsevier.com/about/our-business/policies/article-withdrawal). This article has been withdrawn as part of the withdrawal of the Proceedings of the International Conference on Emerging Trends in Materials Science, Technology and Engineering (ICMSTE2K21). Subsequent to acceptance of these Proceedings papers by the responsible Guest Editors, Dr S. Sakthivel, Dr S. Karthikeyan and Dr I. A. Palani, several serious concerns arose regarding the integrity and veracity of the conference organisation and peer-review process. After a thorough investigation, the peer-review process was confirmed to fall beneath the high standards expected by Materials Today: Proceedings. The veracity of the conference also remains subject to serious doubt and therefore the entire Proceedings has been withdrawn in order to correct the scholarly record.
Preprint
Full-text available
Identification of plant disease is usually done through visual inspection or during laboratory examination which causes delays resulting in yield loss by the time identification is complete. On the other hand, complex deep learning models perform the task with reasonable performance but due to their large size and high computational requirements, they are not suited to mobile and handheld devices. Our proposed approach contributes automated identification of plant diseases which follows a sequence of steps involving pre-processing, segmentation of diseased leaf area, calculation of features based on the Gray-Level Co-occurrence Matrix (GLCM), feature selection and classification. In this study, six color features and twenty-two texture features have been calculated. Support vector machines is used to perform one-vs-one classification of plant disease. The proposed model of disease identification provides an accuracy of 98.79% with a standard deviation of 0.57 on 10-fold cross-validation. The accuracy on a self-collected dataset is 82.47% for disease identification and 91.40% for healthy and diseased classification. The reported performance measures are better or comparable to the existing approaches and highest among the feature-based methods, presenting it as the most suitable method to automated leaf-based plant disease identification. This prototype system can be extended by adding more disease categories or targeting specific crop or disease categories.
Article
Full-text available
Agricultural disease image recognition has an important role to play in the field of intelligent agriculture. Some advanced machine learning methods associated with the development of artificial intelligence technology in recent years, such as deep learning and transfer learning, have begun to be used for the recognition of agricultural diseases. However, the adoption of these methods continues to face a number of important challenges. This paper looks specifically at deep learning and transfer learning and discusses the recent progress in the use of these advanced technologies for agricultural disease image recognition. Analysis and comparison of these two methods reveals that current agricultural disease data resources make transfer learning the better option. The paper then examines the core issues that require further study for research in this domain to continue to progress, such as the construction of image datasets, the selection of big data auxiliary domains and the optimization of the transfer learning method. Creating image datasets obtained under actual cultivation conditions is found to be especially important for the development of practically viable agricultural disease image recognition systems.
Chapter
Smart farming signifies the application of modern information and communication technologies (ICT) into agriculture; plant disease is a threat globally as it reduces the quality of food as well as harmful to health also. To overcome the problem, an automatic detection of diseases is needed that can prevent plants or crops from it by analyzing at early stage and on other hand taking precautions at plantation stage so crop will not have effected by diseases. For easy implementation, deep learning technique is used. DL learns by itself from environment without having human expert. In this survey paper, various deep learning approaches and models been reviewed in recent research papers of plant diseases identification and classification.
Article
Full-text available
Automatic and accurate estimation of disease severity is essential for food security, disease management, and yield loss prediction. Deep learning, the latest breakthrough in computer vision, is promising for fine-grained disease severity classification, as the method avoids the labor-intensive feature engineering and threshold-based segmentation. Using the apple black rot images in the PlantVillage dataset, which are further annotated by botanists with four severity stages as ground truth, a series of deep convolutional neural networks are trained to diagnose the severity of the disease. The performances of shallow networks trained from scratch and deep models fine-tuned by transfer learning are evaluated systemically in this paper. The best model is the deep VGG16 model trained with transfer learning, which yields an overall accuracy of 90.4% on the hold-out test set. The proposed deep learning model may have great potential in disease control for modern agriculture.
Article
Full-text available
The latest generation of convolutional neural networks (CNNs) has achieved impressive results in the field of image classification. This paper is concerned with a new approach to the development of plant disease recognition model, based on leaf image classification, by the use of deep convolutional networks. Novel way of training and the methodology used facilitate a quick and easy system implementation in practice. The developed model is able to recognize 13 different types of plant diseases out of healthy leaves, with the ability to distinguish plant leaves from their surroundings. According to our knowledge, this method for plant disease recognition has been proposed for the first time. All essential steps required for implementing this disease recognition model are fully described throughout the paper, starting from gathering images in order to create a database, assessed by agricultural experts. Caffe, a deep learning framework developed by Berkley Vision and Learning Centre, was used to perform the deep CNN training. The experimental results on the developed model achieved precision between 91% and 98%, for separate class tests, on average 96.3%.
Article
Full-text available
Crop diseases are a major threat to food security, but their rapid identification remains difficult in many parts of the world due to the lack of the necessary infrastructure. The combination of increasing global smartphone penetration and recent advances in computer vision made possible by deep learning has paved the way for smartphone-assisted disease diagnosis. Using a public dataset of 54,306 images of diseased and healthy plant leaves collected under controlled conditions, we train a deep convolutional neural network to identify 14 crop species and 26 diseases (or absence thereof). The trained model achieves an accuracy of 99.35% on a held-out test set, demonstrating the feasibility of this approach. Overall, the approach of training deep learning models on increasingly large and publicly available image datasets presents a clear path toward smartphone-assisted crop disease diagnosis on a massive global scale.
Article
Full-text available
Deep learning allows computational models that are composed of multiple processing layers to learn representations of data with multiple levels of abstraction. These methods have dramatically improved the state-of-the-art in speech recognition, visual object recognition, object detection and many other domains such as drug discovery and genomics. Deep learning discovers intricate structure in large data sets by using the backpropagation algorithm to indicate how a machine should change its internal parameters that are used to compute the representation in each layer from the representation in the previous layer. Deep convolutional nets have brought about breakthroughs in processing images, video, speech and audio, whereas recurrent nets have shone light on sequential data such as text and speech.
Article
Full-text available
When a large feedforward neural network is trained on a small training set, it typically performs poorly on held-out test data. This "overfitting" is greatly reduced by randomly omitting half of the feature detectors on each training case. This prevents complex co-adaptations in which a feature detector is only helpful in the context of several other specific feature detectors. Instead, each neuron learns to detect a feature that is generally helpful for producing the correct answer given the combinatorially large variety of internal contexts in which it must operate. Random "dropout" gives big improvements on many benchmark tasks and sets new records for speech and object recognition.
Conference Paper
Full-text available
The explosion of image data on the Internet has the potential to foster more sophisticated and robust models and algorithms to index, retrieve, organize and interact with images and multimedia data. But exactly how such data can be harnessed and organized remains a critical problem. We introduce here a new database called ldquoImageNetrdquo, a large-scale ontology of images built upon the backbone of the WordNet structure. ImageNet aims to populate the majority of the 80,000 synsets of WordNet with an average of 500-1000 clean and full resolution images. This will result in tens of millions of annotated images organized by the semantic hierarchy of WordNet. This paper offers a detailed analysis of ImageNet in its current state: 12 subtrees with 5247 synsets and 3.2 million images in total. We show that ImageNet is much larger in scale and diversity and much more accurate than the current image datasets. Constructing such a large-scale database is a challenging task. We describe the data collection scheme with Amazon Mechanical Turk. Lastly, we illustrate the usefulness of ImageNet through three simple applications in object recognition, image classification and automatic object clustering. We hope that the scale, accuracy, diversity and hierarchical structure of ImageNet can offer unparalleled opportunities to researchers in the computer vision community and beyond.
Conference Paper
Convolutional neural networks (CNNs) have been widely used in computer vision community, significantly improving the state-of-the-art. In most of the available CNNs, the softmax loss function is used as the supervision signal to train the deep model. In order to enhance the discriminative power of the deeply learned features, this paper proposes a new supervision signal, called center loss, for face recognition task. Specifically, the center loss simultaneously learns a center for deep features of each class and penalizes the distances between the deep features and their corresponding class centers. More importantly, we prove that the proposed center loss function is trainable and easy to optimize in the CNNs. With the joint supervision of softmax loss and center loss, we can train a robust CNNs to obtain the deep features with the two key learning objectives, inter-class dispension and intra-class compactness as much as possible, which are very essential to face recognition. It is encouraging to see that our CNNs (with such joint supervision) achieve the state-of-the-art accuracy on several important face recognition benchmarks, Labeled Faces in the Wild (LFW), YouTube Faces (YTF), and MegaFace Challenge. Especially, our new approach achieves the best results on MegaFace (the largest public domain face benchmark) under the protocol of small training set (contains under 500000 images and under 20000 persons), significantly improving the previous results and setting new state-of-the-art for both face recognition and face verification tasks.
Conference Paper
Learning effective feature representations and similarity measures are crucial to the retrieval performance of a content-based image retrieval (CBIR) system. Despite extensive research efforts for decades, it remains one of the most challenging open problems that considerably hinders the successes of real-world CBIR systems. The key challenge has been attributed to the well-known "semantic gap" issue that exists between low-level image pixels captured by machines and high-level semantic concepts perceived by human. Among various techniques, machine learning has been actively investigated as a possible direction to bridge the semantic gap in the long term. Inspired by recent successes of deep learning techniques for computer vision and other applications, in this paper, we attempt to address an open problem: if deep learning is a hope for bridging the semantic gap in CBIR and how much improvements in CBIR tasks can be achieved by exploring the state-of-the-art deep learning techniques for learning feature representations and similarity measures. Specifically, we investigate a framework of deep learning with application to CBIR tasks with an extensive set of empirical studies by examining a state-of-the-art deep learning method (Convolutional Neural Networks) for CBIR tasks under varied settings. From our empirical studies, we find some encouraging results and summarize some important insights for future research.
Article
Plant disease automatic detection is an important research topic as it has been proved useful in monitoring large crop fields, and thus automatically detects the leaf disease symptoms as soon as they appear in plant leaves. In this paper, a plant disease recognition method is proposed based on plant leaf images. First, the spot is segmented, and the disease feature vector is extracted. Then, the extracted features are provided for the K-nearest-neighbor classifier to recognize the plant diseases. Experimental results show the effectiveness of the proposed approach. © 2015, Pakistan Agricultural Scientists Forum. All rights reserved.
Article
We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions. The method is straightforward to implement and is based an adaptive estimates of lower-order moments of the gradients. The method is computationally efficient, has little memory requirements and is well suited for problems that are large in terms of data and/or parameters. The method is also ap- propriate for non-stationary objectives and problems with very noisy and/or sparse gradients. The method exhibits invariance to diagonal rescaling of the gradients by adapting to the geometry of the objective function. The hyper-parameters have intuitive interpretations and typically require little tuning. Some connections to related algorithms, on which Adam was inspired, are discussed. We also analyze the theoretical convergence properties of the algorithm and provide a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework. We demonstrate that Adam works well in practice when experimentally compared to other stochastic optimization methods.
Article
This monograph provides an overview of general deep learning methodology and its applications to a variety of signal and information processing tasks. The application areas are chosen with the following three criteria in mind: (1) expertise or knowledge of the authors; (2) the application areas that have already been transformed by the successful use of deep learning technology, such as speech recognition and computer vision; and (3) the application areas that have the potential to be impacted significantly by deep learning and that have been experiencing research growth, including natural language and text processing, information retrieval, and multimodal information processing empowered by multi-task deep learning.
Article
Automatic methods for an early detection of plant diseases are vital for precision crop protection. The main contribution of this paper is a procedure for the early detection and differentiation of sugar beet diseases based on Support Vector Machines and spectral vegetation indices. The aim was (I) to discriminate diseased from non-diseased sugar beet leaves, (II) to differentiate between the diseases Cercospora leaf spot, leaf rust and powdery mildew, and (III) to identify diseases even before specific symptoms became visible. Hyperspectral data were recorded from healthy leaves and leaves inoculated with the pathogens Cercospora beticola, Uromyces betae or Erysiphe betae causing Cercospora leaf spot, sugar beet rust and powdery mildew, respectively for a period of 21 days after inoculation. Nine spectral vegetation indices, related to physiological parameters were used as features for an automatic classification. Early differentiation between healthy and inoculated plants as well as among specific diseases can be achieved by a Support Vector Machine with a radial basis function as kernel. The discrimination between healthy sugar beet leaves and diseased leaves resulted in classification accuracies up to 97%. The multiple classification between healthy leaves and leaves with symptoms of the three diseases still achieved an accuracy higher than 86%. Furthermore the potential of presymptomatic detection of the plant diseases was demonstrated. Depending on the type and stage of disease the classification accuracy was between 65% and 90%. (C) 2010 Elsevier B.V. All rights reserved.
Article
A major assumption in many machine learning and data mining algorithms is that the training and future data must be in the same feature space and have the same distribution. However, in many real-world applications, this assumption may not hold. For example, we sometimes have a classification task in one domain of interest, but we only have sufficient training data in another domain of interest, where the latter data may be in a different feature space or follow a different data distribution. In such cases, knowledge transfer, if done successfully, would greatly improve the performance of learning by avoiding much expensive data-labeling efforts. In recent years, transfer learning has emerged as a new learning framework to address this problem. This survey focuses on categorizing and reviewing the current progress on transfer learning for classification, regression, and clustering problems. In this survey, we discuss the relationship between transfer learning and other related machine learning techniques such as domain adaptation, multitask learning and sample selection bias, as well as covariate shift. We also explore some potential future issues in transfer learning research.
Conference Paper
A Research of maize disease image recognition of corn leaf based on image processing and analysis, which is to study diseases of image classification. According to the texture characteristics of corn diseases, it uses YCbCr color space technology to segment disease spot, and uses the cooccurrence matrix spatial gray level layer to extract disease spot texture feature, and uses BP neural network to class the maize disease. Application YCbCr color space technology segmented disease spot, and using the co-occurrence matrix spatial gray level layer extracted disease spot texture feature of using BP neural network, on maize disease classification identification. On VC++ platform, do experiments for the study design recognition algorithm, the experimental results show that the algorithm can effectively identify the disease image, the accuracy was as high as 98% or more, the study provided the theoretical basis to cognition of maize leaf disease. the image re of maize leaf disease image recognition to provide a theoretical basis.
Conference Paper
In this paper, Rough Sets approach has been used to reduce the number of inputs for two neural networks-based applications that are, diagnosing plant diseases and intrusion detection. After the reduction process, and as a result of decreasing the complexity of the classifiers, the results obtained using Multi-Layer Perceptron (MLP) revealed a great deal of classification accuracy without affecting the classification decisions.