Tomato plants' susceptibility to diseases imperils agricultural yields. About 30% of the total crop loss is attributable to plants with disease. Detecting such illnesses in the plant is crucial to avoid significant output losses.This study introduces "data fusion" to enhance disease classification by amalgamating distinct disease-specific traits from leaf halves. Data fusion generates synthetic samples, fortifying a TensorFlow Keras deep learning model using a diverse tomato leaf image dataset. Results illuminate the augmented model's efficacy, particularly for diseases marked by overlapping traits. Enhanced disease recognition accuracy and insights into disease interactions transpire. Evaluation metrics (accuracy 0.95, precision 0.58, recall 0.50, F1 score 0.51) spotlight balanced performance. While attaining commendable accuracy, the intricate precision-recall interplay beckons further examination. In conclusion, data fusion emerges as a promising avenue for refining disease classification, effectively addressing challenges rooted in trait overlap. The integration of TensorFlow Keras underscores the potential for enhancing agricultural practices. Sustained endeavours toward enhanced recall remain pivotal, charting a trajectory for future advancements.
Improved vision-based diagnosis of multi-plant disease using an ensemble of d...IJECEIAES
Farming and plants are crucial parts of the inward economy of a nation, which significantly boosts the economic growth of a country. Preserving plants from several disease infections at their early stage becomes cumbersome due to the absence of efficient diagnosis tools. Diverse difficulties lie in existing methods of plant disease recognition. As a result, developing a rapid and efficient multi-plant disease diagnosis system is a challenging task. At present, deep learning-based methods are frequently utilized for diagnosing plant diseases, which outperformed existing methods with higher efficiency. In order to investigate plant diseases more accurately, this article addresses an efficient hybrid approach using deep learning-based methods. Xception and ResNet50 models were applied for the classification of plant diseases, and these models were merged using the stacking ensemble learning technique to generate a hybrid model. A multi-plant dataset was created using leaf images of four plants: black gram, betel, Malabar spinach, and litchi, which contains nine classes and 44,972 images. Compared to existing individual convolutional neural networks (CNN) models, the proposed hybrid model is more feasible and effective, which acquired 99.20% accuracy. The outcomes and comparison with existing methods represent that the designed method can acquire competitive performance on the multi-plant disease diagnosis tasks.
Optimized deep learning-based dual segmentation framework for diagnosing heal...IAESIJAI
The high disease prevalence in apple farms results in decreased yield and income. This research addresses these issues by integrating internet of things (IoT) applications and deep neural networks to automate disease detection. Existing methods often suffer from high false positives and lack global image similarity. This study proposes a conceptual framework using IoT visual sensors to mitigate apple diseases' severity and presents an intelligent disease detection system. The system employs the augmented Otsu technique for region-aware segmentation and a colour-conversion algorithm for generating feature maps. These maps are input into U-net models, optimized using a genetic algorithm, which results in the generation of suitable masks for all input leaf images. The obtained masks are then used as feature maps to train the convolution neural network (CNN) model for detecting and classifying leaf diseases. Experimental outcomes and comparative assessments demonstrate the proposed scheme's practical utility, yielding high accuracy and low false-positive results in multiclass disease detection tasks.
Early detection of tomato leaf diseases based on deep learning techniquesIAESIJAI
Tomato leaf diseases are a big issue for producers, and finding a single method to combat them is tough. Deep learning techniques, notably convolutional neural networks (CNNs), show promise in recognizing early indicators of illness, which can help producers avoid costly concerns in the future. In this study, we present a CNN-based model for the early identification of tomato leaf diseases to preserve output and boost yield. We used a dataset from the plantvillage database with 11,000 photos from 10 distinct disease categories to train our model. Our CNN was trained on this dataset, and the suggested model obtained an astounding 96% accuracy rate. This shows that our method has the potential to be efficient in detecting tomato leaf diseases early on, therefore assisting producers in managing and reducing disease outbreaks and, as a result, resulting in higher crop yields.
Plant Leaf Diseases Identification in Deep LearningCSEIJJournal
Crop diseases constitute a big threat to plant existence, but their rapid identification remains difficult in
many parts of the planet because of the shortage of the required infrastructure. In computer vision, plant
leaf detection made possible by deep learning has paved the way for smartphone-assisted disease
diagnosis. employing a public dataset of 4,306 images of diseased and healthy plant leaves collected under
controlled conditions, we train a deep convolutional neural network to spot one crop species and
4 diseases (or absence thereof). The trained model achieves an accuracy of 97.35% on a held-out test set,
demonstrating the feasibility of this approach. Overall, the approach of coaching deep learning models on
increasingly large and publicly available image datasets presents a transparent path toward smartphone-
assisted crop disease diagnosis on a large global scale. After the disease is successfully predicted with a
decent confidence level, the corresponding remedy for the disease present is displayed that may be taken as
a cure.
Convolutional neural networks (CNN) trained using deep learning (DL) have advanced dramatically in recent years. Researchers from a variety of fields have been motivated by the success of CNNs in computer vision to develop better CNN models for use in other visually-rich settings. Successes in image classification and research have been achieved in a wide variety of domains throughout the past year. Among the many popularized image classification techniques, the detection of plant leaf diseases has received extensive research. As a result of the nature of the procedure, image quality is often degraded and distortions are introduced during the capturing of the image. In this study, we look into how various CNN models are affected by distortions. Corn-maze leaf photos from the 4,188-image corn or maize leaf Dataset (split into four categories) are under consideration. To evaluate how well they handle noise and blur, researchers have deployed pre-trained deep CNN models like visual geometry group (VGG), InceptionV3, ResNet50, and EfficientNetB0. Classification accuracy and metrics like as recall and f1-score are used to evaluate CNN performance.
Improved vision-based diagnosis of multi-plant disease using an ensemble of d...IJECEIAES
Farming and plants are crucial parts of the inward economy of a nation, which significantly boosts the economic growth of a country. Preserving plants from several disease infections at their early stage becomes cumbersome due to the absence of efficient diagnosis tools. Diverse difficulties lie in existing methods of plant disease recognition. As a result, developing a rapid and efficient multi-plant disease diagnosis system is a challenging task. At present, deep learning-based methods are frequently utilized for diagnosing plant diseases, which outperformed existing methods with higher efficiency. In order to investigate plant diseases more accurately, this article addresses an efficient hybrid approach using deep learning-based methods. Xception and ResNet50 models were applied for the classification of plant diseases, and these models were merged using the stacking ensemble learning technique to generate a hybrid model. A multi-plant dataset was created using leaf images of four plants: black gram, betel, Malabar spinach, and litchi, which contains nine classes and 44,972 images. Compared to existing individual convolutional neural networks (CNN) models, the proposed hybrid model is more feasible and effective, which acquired 99.20% accuracy. The outcomes and comparison with existing methods represent that the designed method can acquire competitive performance on the multi-plant disease diagnosis tasks.
Optimized deep learning-based dual segmentation framework for diagnosing heal...IAESIJAI
The high disease prevalence in apple farms results in decreased yield and income. This research addresses these issues by integrating internet of things (IoT) applications and deep neural networks to automate disease detection. Existing methods often suffer from high false positives and lack global image similarity. This study proposes a conceptual framework using IoT visual sensors to mitigate apple diseases' severity and presents an intelligent disease detection system. The system employs the augmented Otsu technique for region-aware segmentation and a colour-conversion algorithm for generating feature maps. These maps are input into U-net models, optimized using a genetic algorithm, which results in the generation of suitable masks for all input leaf images. The obtained masks are then used as feature maps to train the convolution neural network (CNN) model for detecting and classifying leaf diseases. Experimental outcomes and comparative assessments demonstrate the proposed scheme's practical utility, yielding high accuracy and low false-positive results in multiclass disease detection tasks.
Early detection of tomato leaf diseases based on deep learning techniquesIAESIJAI
Tomato leaf diseases are a big issue for producers, and finding a single method to combat them is tough. Deep learning techniques, notably convolutional neural networks (CNNs), show promise in recognizing early indicators of illness, which can help producers avoid costly concerns in the future. In this study, we present a CNN-based model for the early identification of tomato leaf diseases to preserve output and boost yield. We used a dataset from the plantvillage database with 11,000 photos from 10 distinct disease categories to train our model. Our CNN was trained on this dataset, and the suggested model obtained an astounding 96% accuracy rate. This shows that our method has the potential to be efficient in detecting tomato leaf diseases early on, therefore assisting producers in managing and reducing disease outbreaks and, as a result, resulting in higher crop yields.
Plant Leaf Diseases Identification in Deep LearningCSEIJJournal
Crop diseases constitute a big threat to plant existence, but their rapid identification remains difficult in
many parts of the planet because of the shortage of the required infrastructure. In computer vision, plant
leaf detection made possible by deep learning has paved the way for smartphone-assisted disease
diagnosis. employing a public dataset of 4,306 images of diseased and healthy plant leaves collected under
controlled conditions, we train a deep convolutional neural network to spot one crop species and
4 diseases (or absence thereof). The trained model achieves an accuracy of 97.35% on a held-out test set,
demonstrating the feasibility of this approach. Overall, the approach of coaching deep learning models on
increasingly large and publicly available image datasets presents a transparent path toward smartphone-
assisted crop disease diagnosis on a large global scale. After the disease is successfully predicted with a
decent confidence level, the corresponding remedy for the disease present is displayed that may be taken as
a cure.
Convolutional neural networks (CNN) trained using deep learning (DL) have advanced dramatically in recent years. Researchers from a variety of fields have been motivated by the success of CNNs in computer vision to develop better CNN models for use in other visually-rich settings. Successes in image classification and research have been achieved in a wide variety of domains throughout the past year. Among the many popularized image classification techniques, the detection of plant leaf diseases has received extensive research. As a result of the nature of the procedure, image quality is often degraded and distortions are introduced during the capturing of the image. In this study, we look into how various CNN models are affected by distortions. Corn-maze leaf photos from the 4,188-image corn or maize leaf Dataset (split into four categories) are under consideration. To evaluate how well they handle noise and blur, researchers have deployed pre-trained deep CNN models like visual geometry group (VGG), InceptionV3, ResNet50, and EfficientNetB0. Classification accuracy and metrics like as recall and f1-score are used to evaluate CNN performance.
Diseases in edible and industrial plants remains a major concern, affecting producers and consumers. The problem is further exacerbated as there are different species of plants with a wide variety of diseases that reduce the effectiveness of certain pesticides while increasing our risk of illness. A timely, accurate and automated detection of diseases can be beneficial. Our work focuses on evaluating deep learning (DL) approaches using transfer learning to automatically detect diseases in plants. To enhance the capabilities of our approach, we compiled a novel image dataset containing 87,570 records encompassing 32 different plants and 74 types of diseases. The dataset consists of leaf images from both laboratory setups and cultivation fields, making it more representative. To the best of our knowledge, no such datasets have been used for DL models. Four pretrained computer vision models, namely VGG-16, VGG-19, ResNet-50, and ResNet-101 were evaluated on our dataset. Our experiments demonstrate that both VGG-16 and VGG-19 models proved more efficient, yielding an accuracy of approximately 86% and a f1-score of 87%, as compared to ResNet-50 and ResNet-101. ResNet-50 attains an accuracy and a f1-score of 46.9% and 45.6%, respectively, while ResNet-101 reaches an accuracy of 40.7% and a f1-score of 26.9%.
A deep learning-based approach for early detection of disease in sugarcane pl...IAESIJAI
In many regions of the nation, agriculture serves as the primary industry. The farming environment now faces a number of challenges to farmers. One of the major concerns, and the focus of this research, is disease prediction. A methodology is suggested to automate a process for identifying disease in plant growth and warning farmers in advance so they can take appropriate action. Disease in crop plants has an impact on agricultural production. In this work, a novel DenseNet-support vector machine: explainable artificial intelligence (DNet-SVM: XAI) interpretation that combines a DenseNet with support vector machine (SVM) and local interpretable model-agnostic explanation (LIME) interpretation has been proposed. DNet-SVM: XAI was created by a series of modifications to DenseNet201, including the addition of a support vector machine (SVM) classifier. Prior to using SVM to identify if an image is healthy or un-healthy, images are first feature extracted using a convolution network called DenseNet. In addition to offering a likely explanation for the prediction, the reasoning is carried out utilizing the visual cue produced by the LIME. In light of this, the proposed approach, when paired with its determined interpretability and precision, may successfully assist farmers in the detection of infected plants and recommendation of pesticide for the identified disease.
A study on real time plant disease diagonsis systemIJARIIT
We aim to develop a real time application to the farmers for managing crop diseases. However, disease detection requires
continuous monitoring of experts which might be prohibitively expensive in large farms area. Automatic detection of plant diseases
is an essential research topic as it may prove benefits in monitoring large fields of crops and thus automatically detect the symptoms
of diseases as soon as they appear on plant leaves. Regarding plant disease diagnosis methodologies to detect diseases on crops,
image processing in disease diagnosis and eAGROBOT was studied. This paper is aiming to all are collectively used and formed
semi real time system for a disease diagnosis which uses image processing and data mining concepts to give pesticide
recommendation and pesticide cost estimation system. Thus the android application makes a good foundation for following effective
characteristic parameters for the disease diagnoses and setting up recommender system. The system is to be designed and developed
using Android studio as front-end software and SQLite as back-end software. The pictures and remedial measures of the diseases
were stored in the database and can be retrieved whenever necessary. The challenge is to make the farmers listen to the crop disease
diagnosis system and to get the advice related to the crop diseases. The constraint here is to develop the expert in local languages so
that farmers can operate the ES by themselves and get expert advice from the system.
A comprehensive study on disease risk predictions in machine learning IJECEIAES
Over recent years, multiple disease risk prediction models have been developed. These models use various patient characteristics to estimate the probability of outcomes over a certain period of time and hold the potential to improve decision making and individualize care. Discovering hidden patterns and interactions from medical databases with growing evaluation of the disease prediction model has become crucial. It needs many trials in traditional clinical findings that could complicate disease prediction. A Comprehensive study on different strategies used to predict disease is conferred in this paper. Applying these techniques to healthcare data, has improvement of risk prediction models to find out the patients who would get benefit from disease management programs to reduce hospital readmission and healthcare cost, but the results of these endeavors have been shifted.
A deep learning-based mobile app system for visual identification of tomato p...IJECEIAES
Tomato is one of many horticulture crops in Indonesia which plays a vital role in supplying public food needs. However, tomato is a very susceptible plant to pests and diseases caused by bacteria and fungus. The infected diseases should be isolated as soon as it was detected. Therefore, developing a reliable and fast system is essential for controlling tomato pests and diseases. The deep learning-based application can help to speed up the identification of tomato disease as it can perform direct identification from the image. In this research, EfficientNetB0 was implemented to perform multi-class tomato plant disease classification. The model was then deployed to an android-based application using machine learning (ML) kit library. The proposed system obtained satisfactory results, reaching an average accuracy of 91.4%.
Crop Leaf Disease Diagnosis using Convolutional Neural Networkijtsrd
The major problem that the farmers around the world face is losses, because of pests, disease or a nutrient deficiency. They depend upon the information that they get from the agricultural departments for the diagnosis of plant leaf disease. This process is lengthy and complicated. Here comes a system to help farmers everywhere in the world by automatically detecting plant leaf diseases accurately and within no time. The proposed system is capable of identifying the disease of majorly 5 crops which are corn, sugarcane, wheat, and grape. In this paper, the proposed system uses the Mobile Net model, a type of CNN for classification of leaf disease. Several experiments are performed on the dataset to get the accurate output. This system ensures to give more accurate results than the previous systems. Shivani Machha | Nikita Jadhav | Himali Kasar | Prof. Sumita Chandak ""Crop Leaf Disease Diagnosis using Convolutional Neural Network"" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-4 | Issue-2 , February 2020,
URL: https://www.ijtsrd.com/papers/ijtsrd29952.pdf
Paper Url : https://www.ijtsrd.com/engineering/information-technology/29952/crop-leaf-disease-diagnosis-using-convolutional-neural-network/shivani-machha
Guava fruit disease identification based on improved convolutional neural net...IJECEIAES
Guava fruit cultivation is crucial for Asian economic development, with Indonesia producing 449,970 metric tons between 2022 and 2023. However, technology-based approaches can detect disease symptoms, enhancing production and mitigating economic losses by enhancing quality. In this paper, we introduce an accurate guava fruit disease detection (GFDI) system. It contains the generation of appropriate diseased images and the development of a novel improved convolutional neural network (improvedCNN) that is built depending on the principles of AlexNet. Also, several preprocessing techniques have been used, including data augmentation, contrast enhancement, image resizing, and dataset splitting. The proposed improved-CNN model is trained to identify three common guava fruit diseases using a dataset of 612 images. The experimental findings indicate that the proposed improved-CNN model achieve accuracy 98% for trains and 93% for tests using 0.001 learning rate, the model parameters are decreased by 50,106,831 compared with traditional AlexNet model. The findings of the investigation indicate that the deep learning model improves the accuracy and convergence rate for guava fruit disease prevention.
Food is one of the basic needs of human being. Population is increasing day by day. So, it has become important to grow sufficient amount of crops to feed such a huge population. Agricultural intervention in the livelihood of rural India is about 58%. But with the time passing by, plants are being affected with many kinds of diseases, which cause great harm to the agricultural plant productions. It is very difficult to monitor the plant diseases. It requires tremendous amount of work, expertise in the plant diseases, and also require the excessive processing speed and time. Hence, image processing is used for the detection of plant diseases by just capturing the images of the leaves and comparing it with the data sets available. Latest and fostering technologies like Image processing is used to rectify such issues very effectively. In this project, four consecutive stages are used to discover the type of disease. The four stages include pre-processing, leaf segmentation, feature extraction and classification. This paper aims to support and help the farmers in an efficient way.
Water scarcity is the lack of fresh water resources to meet the standard water demand. There are two type of water scarcity. One is physical. The other is economic water scarcity.
Diseases in edible and industrial plants remains a major concern, affecting producers and consumers. The problem is further exacerbated as there are different species of plants with a wide variety of diseases that reduce the effectiveness of certain pesticides while increasing our risk of illness. A timely, accurate and automated detection of diseases can be beneficial. Our work focuses on evaluating deep learning (DL) approaches using transfer learning to automatically detect diseases in plants. To enhance the capabilities of our approach, we compiled a novel image dataset containing 87,570 records encompassing 32 different plants and 74 types of diseases. The dataset consists of leaf images from both laboratory setups and cultivation fields, making it more representative. To the best of our knowledge, no such datasets have been used for DL models. Four pretrained computer vision models, namely VGG-16, VGG-19, ResNet-50, and ResNet-101 were evaluated on our dataset. Our experiments demonstrate that both VGG-16 and VGG-19 models proved more efficient, yielding an accuracy of approximately 86% and a f1-score of 87%, as compared to ResNet-50 and ResNet-101. ResNet-50 attains an accuracy and a f1-score of 46.9% and 45.6%, respectively, while ResNet-101 reaches an accuracy of 40.7% and a f1-score of 26.9%.
A deep learning-based approach for early detection of disease in sugarcane pl...IAESIJAI
In many regions of the nation, agriculture serves as the primary industry. The farming environment now faces a number of challenges to farmers. One of the major concerns, and the focus of this research, is disease prediction. A methodology is suggested to automate a process for identifying disease in plant growth and warning farmers in advance so they can take appropriate action. Disease in crop plants has an impact on agricultural production. In this work, a novel DenseNet-support vector machine: explainable artificial intelligence (DNet-SVM: XAI) interpretation that combines a DenseNet with support vector machine (SVM) and local interpretable model-agnostic explanation (LIME) interpretation has been proposed. DNet-SVM: XAI was created by a series of modifications to DenseNet201, including the addition of a support vector machine (SVM) classifier. Prior to using SVM to identify if an image is healthy or un-healthy, images are first feature extracted using a convolution network called DenseNet. In addition to offering a likely explanation for the prediction, the reasoning is carried out utilizing the visual cue produced by the LIME. In light of this, the proposed approach, when paired with its determined interpretability and precision, may successfully assist farmers in the detection of infected plants and recommendation of pesticide for the identified disease.
A study on real time plant disease diagonsis systemIJARIIT
We aim to develop a real time application to the farmers for managing crop diseases. However, disease detection requires
continuous monitoring of experts which might be prohibitively expensive in large farms area. Automatic detection of plant diseases
is an essential research topic as it may prove benefits in monitoring large fields of crops and thus automatically detect the symptoms
of diseases as soon as they appear on plant leaves. Regarding plant disease diagnosis methodologies to detect diseases on crops,
image processing in disease diagnosis and eAGROBOT was studied. This paper is aiming to all are collectively used and formed
semi real time system for a disease diagnosis which uses image processing and data mining concepts to give pesticide
recommendation and pesticide cost estimation system. Thus the android application makes a good foundation for following effective
characteristic parameters for the disease diagnoses and setting up recommender system. The system is to be designed and developed
using Android studio as front-end software and SQLite as back-end software. The pictures and remedial measures of the diseases
were stored in the database and can be retrieved whenever necessary. The challenge is to make the farmers listen to the crop disease
diagnosis system and to get the advice related to the crop diseases. The constraint here is to develop the expert in local languages so
that farmers can operate the ES by themselves and get expert advice from the system.
A comprehensive study on disease risk predictions in machine learning IJECEIAES
Over recent years, multiple disease risk prediction models have been developed. These models use various patient characteristics to estimate the probability of outcomes over a certain period of time and hold the potential to improve decision making and individualize care. Discovering hidden patterns and interactions from medical databases with growing evaluation of the disease prediction model has become crucial. It needs many trials in traditional clinical findings that could complicate disease prediction. A Comprehensive study on different strategies used to predict disease is conferred in this paper. Applying these techniques to healthcare data, has improvement of risk prediction models to find out the patients who would get benefit from disease management programs to reduce hospital readmission and healthcare cost, but the results of these endeavors have been shifted.
A deep learning-based mobile app system for visual identification of tomato p...IJECEIAES
Tomato is one of many horticulture crops in Indonesia which plays a vital role in supplying public food needs. However, tomato is a very susceptible plant to pests and diseases caused by bacteria and fungus. The infected diseases should be isolated as soon as it was detected. Therefore, developing a reliable and fast system is essential for controlling tomato pests and diseases. The deep learning-based application can help to speed up the identification of tomato disease as it can perform direct identification from the image. In this research, EfficientNetB0 was implemented to perform multi-class tomato plant disease classification. The model was then deployed to an android-based application using machine learning (ML) kit library. The proposed system obtained satisfactory results, reaching an average accuracy of 91.4%.
Crop Leaf Disease Diagnosis using Convolutional Neural Networkijtsrd
The major problem that the farmers around the world face is losses, because of pests, disease or a nutrient deficiency. They depend upon the information that they get from the agricultural departments for the diagnosis of plant leaf disease. This process is lengthy and complicated. Here comes a system to help farmers everywhere in the world by automatically detecting plant leaf diseases accurately and within no time. The proposed system is capable of identifying the disease of majorly 5 crops which are corn, sugarcane, wheat, and grape. In this paper, the proposed system uses the Mobile Net model, a type of CNN for classification of leaf disease. Several experiments are performed on the dataset to get the accurate output. This system ensures to give more accurate results than the previous systems. Shivani Machha | Nikita Jadhav | Himali Kasar | Prof. Sumita Chandak ""Crop Leaf Disease Diagnosis using Convolutional Neural Network"" Published in International Journal of Trend in Scientific Research and Development (ijtsrd), ISSN: 2456-6470, Volume-4 | Issue-2 , February 2020,
URL: https://www.ijtsrd.com/papers/ijtsrd29952.pdf
Paper Url : https://www.ijtsrd.com/engineering/information-technology/29952/crop-leaf-disease-diagnosis-using-convolutional-neural-network/shivani-machha
Guava fruit disease identification based on improved convolutional neural net...IJECEIAES
Guava fruit cultivation is crucial for Asian economic development, with Indonesia producing 449,970 metric tons between 2022 and 2023. However, technology-based approaches can detect disease symptoms, enhancing production and mitigating economic losses by enhancing quality. In this paper, we introduce an accurate guava fruit disease detection (GFDI) system. It contains the generation of appropriate diseased images and the development of a novel improved convolutional neural network (improvedCNN) that is built depending on the principles of AlexNet. Also, several preprocessing techniques have been used, including data augmentation, contrast enhancement, image resizing, and dataset splitting. The proposed improved-CNN model is trained to identify three common guava fruit diseases using a dataset of 612 images. The experimental findings indicate that the proposed improved-CNN model achieve accuracy 98% for trains and 93% for tests using 0.001 learning rate, the model parameters are decreased by 50,106,831 compared with traditional AlexNet model. The findings of the investigation indicate that the deep learning model improves the accuracy and convergence rate for guava fruit disease prevention.
Food is one of the basic needs of human being. Population is increasing day by day. So, it has become important to grow sufficient amount of crops to feed such a huge population. Agricultural intervention in the livelihood of rural India is about 58%. But with the time passing by, plants are being affected with many kinds of diseases, which cause great harm to the agricultural plant productions. It is very difficult to monitor the plant diseases. It requires tremendous amount of work, expertise in the plant diseases, and also require the excessive processing speed and time. Hence, image processing is used for the detection of plant diseases by just capturing the images of the leaves and comparing it with the data sets available. Latest and fostering technologies like Image processing is used to rectify such issues very effectively. In this project, four consecutive stages are used to discover the type of disease. The four stages include pre-processing, leaf segmentation, feature extraction and classification. This paper aims to support and help the farmers in an efficient way.
Water scarcity is the lack of fresh water resources to meet the standard water demand. There are two type of water scarcity. One is physical. The other is economic water scarcity.
Automobile Management System Project Report.pdfKamal Acharya
The proposed project is developed to manage the automobile in the automobile dealer company. The main module in this project is login, automobile management, customer management, sales, complaints and reports. The first module is the login. The automobile showroom owner should login to the project for usage. The username and password are verified and if it is correct, next form opens. If the username and password are not correct, it shows the error message.
When a customer search for a automobile, if the automobile is available, they will be taken to a page that shows the details of the automobile including automobile name, automobile ID, quantity, price etc. “Automobile Management System” is useful for maintaining automobiles, customers effectively and hence helps for establishing good relation between customer and automobile organization. It contains various customized modules for effectively maintaining automobiles and stock information accurately and safely.
When the automobile is sold to the customer, stock will be reduced automatically. When a new purchase is made, stock will be increased automatically. While selecting automobiles for sale, the proposed software will automatically check for total number of available stock of that particular item, if the total stock of that particular item is less than 5, software will notify the user to purchase the particular item.
Also when the user tries to sale items which are not in stock, the system will prompt the user that the stock is not enough. Customers of this system can search for a automobile; can purchase a automobile easily by selecting fast. On the other hand the stock of automobiles can be maintained perfectly by the automobile shop manager overcoming the drawbacks of existing system.
Explore the innovative world of trenchless pipe repair with our comprehensive guide, "The Benefits and Techniques of Trenchless Pipe Repair." This document delves into the modern methods of repairing underground pipes without the need for extensive excavation, highlighting the numerous advantages and the latest techniques used in the industry.
Learn about the cost savings, reduced environmental impact, and minimal disruption associated with trenchless technology. Discover detailed explanations of popular techniques such as pipe bursting, cured-in-place pipe (CIPP) lining, and directional drilling. Understand how these methods can be applied to various types of infrastructure, from residential plumbing to large-scale municipal systems.
Ideal for homeowners, contractors, engineers, and anyone interested in modern plumbing solutions, this guide provides valuable insights into why trenchless pipe repair is becoming the preferred choice for pipe rehabilitation. Stay informed about the latest advancements and best practices in the field.
Vaccine management system project report documentation..pdfKamal Acharya
The Division of Vaccine and Immunization is facing increasing difficulty monitoring vaccines and other commodities distribution once they have been distributed from the national stores. With the introduction of new vaccines, more challenges have been anticipated with this additions posing serious threat to the already over strained vaccine supply chain system in Kenya.
CFD Simulation of By-pass Flow in a HRSG module by R&R Consult.pptxR&R Consult
CFD analysis is incredibly effective at solving mysteries and improving the performance of complex systems!
Here's a great example: At a large natural gas-fired power plant, where they use waste heat to generate steam and energy, they were puzzled that their boiler wasn't producing as much steam as expected.
R&R and Tetra Engineering Group Inc. were asked to solve the issue with reduced steam production.
An inspection had shown that a significant amount of hot flue gas was bypassing the boiler tubes, where the heat was supposed to be transferred.
R&R Consult conducted a CFD analysis, which revealed that 6.3% of the flue gas was bypassing the boiler tubes without transferring heat. The analysis also showed that the flue gas was instead being directed along the sides of the boiler and between the modules that were supposed to capture the heat. This was the cause of the reduced performance.
Based on our results, Tetra Engineering installed covering plates to reduce the bypass flow. This improved the boiler's performance and increased electricity production.
It is always satisfying when we can help solve complex challenges like this. Do your systems also need a check-up or optimization? Give us a call!
Work done in cooperation with James Malloy and David Moelling from Tetra Engineering.
More examples of our work https://www.r-r-consult.dk/en/cases-en/
Quality defects in TMT Bars, Possible causes and Potential Solutions.PrashantGoswami42
Maintaining high-quality standards in the production of TMT bars is crucial for ensuring structural integrity in construction. Addressing common defects through careful monitoring, standardized processes, and advanced technology can significantly improve the quality of TMT bars. Continuous training and adherence to quality control measures will also play a pivotal role in minimizing these defects.
NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...Amil Baba Dawood bangali
Contact with Dawood Bhai Just call on +92322-6382012 and we'll help you. We'll solve all your problems within 12 to 24 hours and with 101% guarantee and with astrology systematic. If you want to take any personal or professional advice then also you can call us on +92322-6382012 , ONLINE LOVE PROBLEM & Other all types of Daily Life Problem's.Then CALL or WHATSAPP us on +92322-6382012 and Get all these problems solutions here by Amil Baba DAWOOD BANGALI
#vashikaranspecialist #astrologer #palmistry #amliyaat #taweez #manpasandshadi #horoscope #spiritual #lovelife #lovespell #marriagespell#aamilbabainpakistan #amilbabainkarachi #powerfullblackmagicspell #kalajadumantarspecialist #realamilbaba #AmilbabainPakistan #astrologerincanada #astrologerindubai #lovespellsmaster #kalajaduspecialist #lovespellsthatwork #aamilbabainlahore#blackmagicformarriage #aamilbaba #kalajadu #kalailam #taweez #wazifaexpert #jadumantar #vashikaranspecialist #astrologer #palmistry #amliyaat #taweez #manpasandshadi #horoscope #spiritual #lovelife #lovespell #marriagespell#aamilbabainpakistan #amilbabainkarachi #powerfullblackmagicspell #kalajadumantarspecialist #realamilbaba #AmilbabainPakistan #astrologerincanada #astrologerindubai #lovespellsmaster #kalajaduspecialist #lovespellsthatwork #aamilbabainlahore #blackmagicforlove #blackmagicformarriage #aamilbaba #kalajadu #kalailam #taweez #wazifaexpert #jadumantar #vashikaranspecialist #astrologer #palmistry #amliyaat #taweez #manpasandshadi #horoscope #spiritual #lovelife #lovespell #marriagespell#aamilbabainpakistan #amilbabainkarachi #powerfullblackmagicspell #kalajadumantarspecialist #realamilbaba #AmilbabainPakistan #astrologerincanada #astrologerindubai #lovespellsmaster #kalajaduspecialist #lovespellsthatwork #aamilbabainlahore #Amilbabainuk #amilbabainspain #amilbabaindubai #Amilbabainnorway #amilbabainkrachi #amilbabainlahore #amilbabaingujranwalan #amilbabainislamabad
Sachpazis:Terzaghi Bearing Capacity Estimation in simple terms with Calculati...Dr.Costas Sachpazis
Terzaghi's soil bearing capacity theory, developed by Karl Terzaghi, is a fundamental principle in geotechnical engineering used to determine the bearing capacity of shallow foundations. This theory provides a method to calculate the ultimate bearing capacity of soil, which is the maximum load per unit area that the soil can support without undergoing shear failure. The Calculation HTML Code included.
Final project report on grocery store management system..pdfKamal Acharya
In today’s fast-changing business environment, it’s extremely important to be able to respond to client needs in the most effective and timely manner. If your customers wish to see your business online and have instant access to your products or services.
Online Grocery Store is an e-commerce website, which retails various grocery products. This project allows viewing various products available enables registered users to purchase desired products instantly using Paytm, UPI payment processor (Instant Pay) and also can place order by using Cash on Delivery (Pay Later) option. This project provides an easy access to Administrators and Managers to view orders placed using Pay Later and Instant Pay options.
In order to develop an e-commerce website, a number of Technologies must be studied and understood. These include multi-tiered architecture, server and client-side scripting techniques, implementation technologies, programming language (such as PHP, HTML, CSS, JavaScript) and MySQL relational databases. This is a project with the objective to develop a basic website where a consumer is provided with a shopping cart website and also to know about the technologies used to develop such a website.
This document will discuss each of the underlying technologies to create and implement an e- commerce website.
Tomato Disease Fusion and Classification using Deep Learning
1. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
Bibhu Dash et al: ISTECH, ITEORY, CSITAI -2023
pp. 31-43, 2023. IJCI – 2023 DOI:10.5121/ijci.2023.120703
TOMATO DISEASE FUSION AND CLASSIFICATION
USING DEEP LEARNING
Patrick Ansah1
, Sumit Kumar Tetarave1
, Ezhil Kalaimannan2
and Caroline John2
1
School of Computer Applications, Kalinga Institute of Industrial Technology, India
2
Department of Computer Science,University of West Florida, FL 32514, USA
ABSTRACT
Tomato plants' susceptibility to diseases imperils agricultural yields. About 30% of the total crop loss is
attributable to plants with disease. Detecting such illnesses in the plant is crucial to avoid significant
output losses.This study introduces "data fusion" to enhance disease classification by amalgamating
distinct disease-specific traits from leaf halves. Data fusion generates synthetic samples, fortifying a
TensorFlow Keras deep learning model using a diverse tomato leaf image dataset. Results illuminate the
augmented model's efficacy, particularly for diseases marked by overlapping traits. Enhanced disease
recognition accuracy and insights into disease interactions transpire. Evaluation metrics (accuracy 0.95,
precision 0.58, recall 0.50, F1 score 0.51) spotlight balanced performance. While attaining commendable
accuracy, the intricate precision-recall interplay beckons further examination. In conclusion, data fusion
emerges as a promising avenue for refining disease classification, effectively addressing challenges rooted
in trait overlap. The integration of TensorFlow Keras underscores the potential for enhancing agricultural
practices. Sustained endeavours toward enhanced recall remain pivotal, charting a trajectory for future
advancements.
KEYWORDS
Disease Fusion, Deep Learning Classification,Tomato Leaf Diseases, TensorFlow Keras, Disease
Recognition
1. INTRODUCTION
Tomato (Solanum lycopersicum), one of the most extensively cultivated and economically
significant crops worldwide, faces numerous diseases that adversely affect its leaves, ultimately
compromising its overall health and productivity. These diseases often manifest as symptoms
such as discoloration, lesions, spots, and leaf deformities. Various pathogens, including fungi,
bacteria, viruses, and environmental factors, can trigger these ailments. Examples of tomato
diseases encompass bacterial spots, early blight, target spot, late blight, leaf mold, yellow leaf
curl virus, septoria leaf spot, and spotted spider mite. To ensure sustainable tomato production,
comprehending and effectively managing these diseases is crucial, given the pivotal role that
tomato leaves play in photosynthesis and the overall vitality of the plant.
Image processing involves the intricate manipulation and analysis of visual data captured in
images. Various algorithms and techniques are employed to enhance, transform, or extract
valuable insights from digital images. This interdisciplinary field integrates aspects of computer
science, mathematics, and engineering to process images for diverse applications, including
image enhancement, object recognition, pattern discovery, analysis of plant disease imagery, and
medical imaging. In disease analysis, image processing enables efficient analysis and
quantification of various elements, such as disease symptoms, patterns, and their correlations
with environmental factors. Image processing aims to enhance and interpret visual input using
2. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
32
computational methods, enabling computers to perceive and comprehend images analogous to
human vision.
There is always a necessity to use scientific techniques because it is exceedingly challenging and
ineffective to diagnose diseases by eye [1]. Farmers in the majority of our native environments
are compelled to think creatively about traits or characteristics that have evolved, and these
enable them to develop appropriate mechanisms that may help in the eradication of underlying
diseases or pest attacks [2]. Formal training is necessary to give people the scientific information
to understand this issue wisely and generate cutting-edge solutions to remove it [3].
Although several machine learning models are utilized for the recognition and classification of
images, the potential of conventional image recognition algorithms is often only partially
realized. This field of Deep Learning study has lots of potential in terms of increased accuracy,
especially given the expanding growth of Deep Learning technology, which has benefitted the
agricultural industry. Precision agriculture regularly uses image processing. Agriculture-related
image processing was covered in-depth in the literature. Image processing is employed in various
applications, including identifying, quantifying, and categorizing plant illnesses [4] and
phenotyping plant disease signs [5].
Deep learning, being a prolific tool, has been frequently utilized in plant disease detection, with a
specific emphasis on its application in diagnosing diseases affecting tomato plants. Mohanty et al.
[6] used deep learning to detect illnesses in the leaves of several plants. The paper's approach is
tailored to address the detection of prevalent tomato plant diseases, including bacterial leaf spot,
septoria leaf spot, and numerous others. It can classify input leaf images into specific disease
categories or ascertain their health status. A dataset derived from a subset of PlantVillage [7] was
utilized for evaluation, encompassing 15 directories pertaining to three distinct crop varieties.
The subgroup includes around 16,024 images of tomato leaf diseases. Similarly, authors in [8]
augmented the data, which consists of 18 diseases of tomatoes.
Moreover, Prajwala et al. [9] worked on unstructured images and classified them. They
performed experiments using AlexNet [10] and GoogleNet [11], and the best results were
obtained during the use of LeNet architecture [12]. The Gabor wavelet transformation technique
has been used to extract and identify tomato diseases [13]. Ashqar et al. [14] performed a
controlled laboratory environment to produce images of healthy and unhealthy tomato leaves.
Llorca et al. [15] collected the images from Google Images to identify the different diseases.
Recently, scientists developed several customized convolutional neural network models and
transfer learning (TL)-based models to identify tomato leaf disease [16-19].
Reviewing previous research has overwhelmingly demonstrated that the majority of deep
learning studies focused on predicting tomato diseases have typically centered on predicting a
single disease in each test. We found that the monopoly detection mechanisms mostly predicted
different diseases in the presence of another disease on the same tomato leaf. We proposed an
augmentation process on the PlantVillage dataset to confuse the existing prediction tools. We
propose a four-way method for this work: Data Acquisition, Pre-processing, Data Splitting and
Fusing, and Classification. The final stage of the classification utilizes TensorFlow's Keras, which
has an intuitive and powerful framework commonly used to build classification models.
The rest of the research paper is structured as follows: Section 2 focuses on the existing research
in the relevant field. Section 3 explains the methods for acquiring the required findings alongside
the model and proposed technique. Section 4 presents the results and examines the suggested
methodology. The conclusion and outlines of the future research directions are presented in
Section 5.
3. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
33
2. LITERATURE SURVEY
Plant leaf disease detection constitutes a significant research domain, where the amalgamation of
image processing and deep learning techniques has resulted in precise classification. This article
delves into prevalent methods incorporated within the literature in this domain.
In the paper [8], the authors used a dataset that contained 18 diseases of tomatoes. They used
augmentation to increase the dataset from 13,112 to 41,263. They employed five types of
Convolutional Neural Networks for classification based on the training they are
DenseNet_Xception, Xception, Resnet50, MobileNet, and ShuffleNet. They comprise layers that
perform convolutional operations to extract local features from the input data. The accuracy
during training was obtained as 97.10%, 93.10%, 86.56%, 80.11%, and 83.68% from
DenseNet_Xception, Xception, Resnet50, MobileNet, and ShuffleNet, respectively. The best
recognition accuracy of DenseNet_Xception is 97.10%, though the parameters for this method
were the most in number. ShuffleNet employed the least number of parameters but recorded an
accuracy of 83.68%. ShuffleNet was trained using an augmented data set with optimized
parameters and performed well during training and validation. The effect of epoch on the
accuracy consistently rises and converges at 1 without any significant drops, typically indicating
that the neural network has learned the training data exceptionally well. The model achieved
learning success by keeping the Converging about the 1 point through to 1000, 2000, and 3000
epochs. If accuracy increases, a loss is expected to depreciate in the manner of optimum training.
In the paper [9], the authors began with Data Acquisition, Data preprocessing, and classification
using Convolutional Neural Networks (CNN) to develop the model that worked on unstructured
images and converted them to corresponding classification output labels. The model was trained
using 10, 20, and 30 epochs, which are mentioned in Table 1. Validation was performed using a
Confusion Matrix. From the outcome, the observation was that, as the number of training
processes (epochs) increased, so did the accuracy, prediction, recall, and consequently, the F1-
Score.
Table 1. Evaluation of epochs
The methodology detailed in the paper [13] presented how the Gabor wavelet transformation
technique was employed to extract distinctive features that helped to identify diseases in tomato
leaves. Subsequently, these extracted features are input into a Support Vector Machine (SVM)
classifier for training, enabling the determination of the specific disease affecting the tomato
leaves. Before the feature extraction, the preprocessing stage involves image resizing, noise
reduction, and background elimination tasks. The research used the Gabor transformation to
capture textual patterns inherent in the affected leaves and extract relevant features. Disease
classification conducted using SVM with varying kernel functions was performed. There was
cross-validation on performance evaluation. TheROC curve of SVM using the invmult kernel
produced an AUC of 0.90705, while the one using the Laplacian kernel produced an AUC of
0.99679. Experimental results indicate an impressive accuracy rate of 99.5% achieved by the
proposed system.
4. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
34
However, it is necessary to note the utilization of Gabor transformation for feature extraction
comes with the limitation of computational intensiveness. In the paper [14], the authors employed
a dataset comprising 9000 images of infected and healthy tomato leaves. All images were
produced within a controlled laboratory environment. This dataset, obtained from the
PlantVillage repository, was utilized for the purpose of categorizing five specific diseases: leaf
curl, bacterial spot, septoria leafspot, early blight, and leaf mold. A comprehensive color model
was utilized for disease spot classification to facilitate data visualization, while a grayscale model
was employed to capture the underlying leaf shapes and visual disease patterns. The results
revealed that the full-color model achieved superior accuracy compared to the grayscale model.
Notably, the captured results were obtained under precisely controlled conditions within the
PlantVillage dataset, thereby introducing a potential limitation to the model's applicability in a
more diverse setting.
In the paper [15], the authors used a dataset comprising 2,779 images from Google Images. These
images encompassed various instances of hornworms, powdery mildew, cutworms, early blight,
and whiteflies. Notably, each disease category contained 550 images. This quantity is considered
limited for robust training, potentially giving rise to overfitting concerns. The author employed
data augmentation techniques to mitigate the overfitting challenge, including vertical flips and
random scaling. These techniques introduce diversity and variability into the dataset, contributing
to a more generalized and resilient model.
While Convolutional Neural Networks (CNNs) are well suited for deciphering image content, it
is worth noting that training a CNN from scratch demands substantial computational resources
and a vast dataset. The author opted for a transfer learning approach, utilizing the Google
Inception model as a foundation. By leveraging pre-trained weights and features from the
Tensorflow Inception V3 model, the author was able to capitalize on its learned representations
and optimize the training process, effectively sidestepping the need for extensive data and
computational power after the model achieved an accuracy of 88.90%.
The dataset for the paper [16] was obtained from Ehime University in Matsuyama. The scientists
developed several simultaneous convolutional neural networks with various topologies to identify
tomato leaf disease. To significantly improve the network's performance, they used the activation
layers Swish, LeakyReLU-Swish, ReLU-Swish, Elu-Swish, and ClippedReLU-Swish in addition
to the Batch Normalization-Instance Normalisation layer. That allowed them to achieve
classification accuracy of over 99.0% with training datasets, 97.5% with validation datasets, and
98.0% with testing datasets. Although different performance metrics were observed, none of the
suggested networks overfit the validation dataset.
They also employed a variety of methods to visualize network performance. It showed how the
networks (Network 1, Network 2, Network 3, Network 4, Network 5) learn from the training
dataset and could show infected leaf areas with high confidence scores under actual
circumstances. In terms of network stability and illness location visualization, Network 1
performed the best. The shortcomings of Networks 4 and 5 in predicting the Healthy class could
be resolved by computing, summarising, and rating the output of several parallel convolutional
neural networks [16].
Reference [17] introduces a study that put forward a custom convolutional neural network (CNN)
model with a lightweight architecture and employed Transfer Learning models VGG-16 and
VGG-19 to classify tomato leaf diseases. This research utilized eleven classes, with one type
dedicated to healthy leaves, to replicate various tomato leaf disease scenarios. Furthermore, the
study conducted an ablation study to pinpoint the most influential parameters for the proposed
model. Moreover, evaluation metrics were used to analyze and compare the performance of the
5. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
35
proposed model with the TL-based model. By applying data augmentation techniques, the
proposed model achieved the highest accuracy and recall of 95.00% among all the models.
Finally, the best-performing model was utilized to construct a Web-based and Android-based
end-to-end (E2E) system for tomato cultivators to classify tomato leaf disease.
3. PROPOSED METHODOLOGIES
3.1. SYSTEM OVERVIEW
This research aims to create a model using deep learning that can predict several diseases of
tomatoes on the same leaf. Therefore, splitting and fusing distinct diseased leaves would be
beneficial. The proposed Split and Fuse model seeks to employ the procedures shown in Figure 1.
Figure1. The proposed procedure from data collection to evaluation of the results.
3.2. Data Collection
The dataset comprises images of tomato leaf diseases sourced from the Plant Village repository
[7]. With a collection of about 16,024 images, it encompasses ten distinct classes. These classes
encompass a comprehensive range of leaf diseases that have the potential to impact tomato crops.
Some of these diseases are Bacterial spot, Early blight, Late blight, Leaf mold, septoria leaf spot,
Target spot, etc.
3.3. Data Pre-processing
The dataset we obtained featured images that exhibited minimal noise, obviating the requirement
for noise removal as a preliminary procedure. The images were resized to dimensions of 255 ×
255 to accelerate the training phase and render model training computationally viable. The neural
representation of resizing and rescaling is shown in Figure 2.
6. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
36
Figure2. The process or pre-processing of the proposed method
Deep learning models can effectively train on small-sized images; therefore, reducing the size of
the input images would contribute to the effective training of the model. However, it is important
to note that while the images are scaled down slightly, the essential features and patterns
necessary for the disease classifications get preserved due to the minimal reduction in size. That
enables the training process to be more resource-effective without significantly sacrificing the
accuracy of the model's predictions.
3.4. Data Splitting and Fusing
Conventional data augmentation techniques aim at diversifying training data by applying
transformations to existing samples. Commonly used augmentation techniques include rotations,
flips, translations, changes in brightness and contrast, and more. These transformations help the
model become more robust by exposing it to variations of the same data. The method needs to
incorporate a seasoned approach. The brain behind this work is to transform the data by splitting
each image into two halves from a particular class and fusing each half with another disease
image from a different class without duplication to maintain uniqueness. Figure 3 shows the
process of splitting and fusing the images.
Figure 3. The process of splitting and fusing the images
Through data splitting and fusing, an image from a particular class undergoes division into halves
and subsequent fusion with corresponding halves sourced from different ones. The overarching
objective of this technique is to replicate intricate interactions across diverse disease classes,
thereby generating fresh instances that might not naturally arise. This approach seeks to facilitate
7. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
37
the model's training in discerning nuanced disease characteristics, ultimately bolstering its
capacity to generalize effectively to real-world scenarios.
With a focus on maintaining distinctiveness, the dataset was constructed by implementing the
split-and-fuse method on a selected number of classes. This undertaking resulted in the creating
of a total of 20 new sub-classes for the designated activity. An illustrative example is presented in
Figure 4, showcasing the fusion of half of an image depicting Tomato Leaf Mold. The Tomato
Leaf Mold class is with half of the Healthy class. Additionally, it is noteworthy that a hybrid
image featuring half of the septoria leaf spot has been seamlessly integrated with late blight, as
exemplified.
Figure 4. Fused diseases from different classes
3.5. Augmentation
In the context of Convolutional Neural Networks within the domain of Deep Learning, achieving
a more robust outcome is mapped to the availability of a substantial volume of data [8]. Even
with the possession of a dataset comprising 15,000 instances, the strategic implementation of
methodologies aimed at data augmentation emerges as a pivotal avenue for enhancing the quality
of this study. Data augmentation represents a canonical strategy employed to amplify the expanse
of the dataset corpus, concurrently affording a mitigation mechanism against the potential pitfalls
of overfitting [16]. The significance of this approach lies in its ability to integrate modified
images into the training dataset, utilizing a variety of techniques like image flipping, rotation,
color manipulation, stochastic cropping, and other methods, as cited in reference [17]. Figure 5
illustrates how images are flipped and rotated.
Figure 5. The proposed method to flip and rotate the images
We leveraged the Keras API within TensorFlow. Through this approach, we established a
preprocessing pipeline that takes care of the resizing and rescaling input images, ensuring their
proper formatting and scaling before their inclusion in the neural network model for training.
Keras, an open-source high-level neural network API has been designed for creating effective
8. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
38
deep learning models. The pixel values were also rescaled by dividing them by 255, bringing
them into the range of [0,1] for normalization.
The neural network arrangement for resizing and rescaling is shown in Figure 6.
Figure 6. The resize and rescale process
4. CLASSIFICATION MODEL
We adopt the TensorFlow Keras API for the image classification. The arrangement of layers of
the Convolutional Neural Networks used for creating the proposed model is described as follows.
4.1. Initial Convolutional Layers
The model's inception involves a convolutional layer that incorporates 32 filters. Each filter is
configured with a 3x3 kernel and operates with the rectified linear unit (ReLU) activation
function.It's one of artificial neural networks' most commonly used activation functions,
especially in deep learning architectures. ReLU is a non-linear activation function that adds non-
linearity to the network, enabling it to learn intricate patterns and representations from the
input.After each convolution, a max-pooling layer with a 2x2 window size is employed to down-
sample the spatial dimensions of the feature maps.
4.2. Intermediate Convolutional Layers
Following the initial convolutional layer, a second convolutional layer is adopted with 64 filters
utilizing a 3x3 kernel and ReLU activation. Another max-pooling layer, again with a 2x2 window
size, is then applied further to reduce the dimensions of the feature maps. A final convolutional
layer utilizes 64 filters with a 3x3 kernel and ReLU activation before a last pooling operation.
4.3. Flattening and Dense Layers
After the convolutional layers, the resultant feature maps get flattened into a 1D vector. A fully
connected dense layer comprising 64 units and the ReLU activation function gets utilized to
capture higher-level patterns and representations.
9. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
39
4.4. Output Layer
The model concludes with a dense layer with some units equal to 20 classes. The Softmax
activation function is employed, which permits the model to yield class probabilities as outputs,
i.e., it is to produce the final output or predictions of a neural network. We also incorporated
dense layers to act as intermediate layers to perform feature extraction and transformation.The
purpose of feature extraction and transformation by the dense layer in the network is to convert
raw input data into a more meaningful and representative form in the form of reduced dimension.
In our case, it reduces the data input. The dense layer, therefore, enables our neural network to
learn and represent relevant information from the input data, facilitating its ability to make
accurate predictions or classifications.
This proposed architecture, constructed through the Sequential API, is tailored to accommodate
input data with dimensions of (batch size, image size, image size, and channels). The proposed
model leverages convolutional layers to extract hierarchical image features and subsequently
applies fully connected layers to enable accurate classification across multiple classes. In
addition, to minimize the loss function and enhance the model's performance on the training data,
three sets of epochs were executed to facilitate the model's training.
It anticipates that this model will significantly contribute to accurately categorizing images by the
specified research objectives.
5. EVALUATION
A set of quantitative parameters comprising accuracy, precision, recall, and F1-score are utilized
to gauge how effective the proposed model is. The findings have been presented in Table 2,
showcasing the highest values of these quantitative metrics achieved up to their corresponding
epoch numbers. Both tests with four and ten epochs exhibit the same accuracy of 0.93. These
observations suggest that the model's accuracy stabilizes after four epochs, and additional epochs
do not substantially improve overall correct predictions on the dataset.
Table 2. Evaluation of parameters
Figure 7 shows the abovementioned phenomenon, where the model's predictions deviated from
expected outcomes when tested against datasets. These two parameters significantly contribute to
the weakness observed in the Prediction and Recall metrics provided by the confusion matrix.
Hence, the model requires enhancement to attain higher accuracy, a pivotal aspect in determining
its utility in practical applications and considering Figure 7 when the model created from the
epoch of four predicted otherwise due to low precision.
10. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
40
Figure 7. Incorrect predictions with low confidence percentages
Epoch graphs with accuracy and validation are significantly meant to provide insights into a
model's training progress and help identify overfitting. With four epochs, the recall is 0.05. That
means that the model correctly identified only 5% of the positive instances in the dataset as
positive. With ten epochs, the recall increased to 0.22. That indicates the model now captures a
large portion of actual positive instances in its predictions. The influences of these instances are
seen in the figure below, where training and validation struggle to exhibit their dominance in the
model creation regarding accuracy and loss. Figure 8 depicts the training and validation
accuracies of the model, which shows that ten epochs properly work when compared to the epoch
of 4.
Figure 8. Training and validation accuracies withepoch of 4 and epoch of 10
Considering four epochs in Figure 8, the F1-score is 0.07. The F1-Score is calculated as the
harmonic mean of precision and recall. It is valuable for evaluating a model's ability to correctly
classify positive instances while minimizing false positives and false negatives. A low F1 score
indicates how the model strives to balance false positives and negatives. On the other hand, the
F1-score has increased to 0.23 with ten epochs. That suggests that the model's precision and
recall have improved in a way that positively affects the overall balance captured by the F1-score.
That balance was realized during its test using a dataset, as shown in Figure 9.
11. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
41
Figure 9. Correct Prediction vs Wrong Prediction
Figure 9 shows an accurate prediction after the model is provided with a dataset at a confidence
of 80.82%. The same model made incorrect predictions with a confidence level of 20.88%. When
supplied with the Mosaic virus and Early blight, the model erroneously predicted health,
suggesting the need for improvement. Moreover, with an epoch count of 50, the model's overall
effectiveness improved compared to epochs of 4 and 10. The model trained for 50 epochs
demonstrated a better capability to capture a more significant portion of actual positive instances
in its predictions.
While accuracy remains consistent between the 4-epoch and 10-epoch models, the 50-epoch
model achieves an improved balance between precision and recall. That implies that extended
training enabled the model to become more accurate, particularly in correctly classifying positive
instances, resulting in a more balanced and higher-performing model overall. Figure 14 (left side)
illustrates training visualization at 50 epochs, demonstrating progressive enhancement compared
to 4 and 10 epochs. On the right side of Figure 10, the model achieves highly accurate predictions
with confidence levels as high as 99.82% when provided with a test dataset. Nevertheless, the
model exhibited enormous weaknesses in other predictions, e.g., in Figure 11.
Figure 10. An accurate Prediction with high confidence
12. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
42
Figure 11.An example of incorrect predictions
Table 3. Details of 50 epochs
Table 3 shows the recorded validation loss value of 1.0999 and a validation accuracy of 0.65,
providing insights into the model's capacity to generalize its acquired knowledge to previously
unseen data. The loss value indicates the model's proficiency in minimizing errors during
validation, while the validation accuracy affirms its ability to sustain high performance beyond
the training dataset. A relatively elevated count of True Negatives (1391) signifies the model's
competence in accurately identifying the absence of the target class. Consequently, this
proficiency also increases the likelihood of generating inaccurate predictions. These metrics
underscore the positive influence of the model's architecture and training process on its
performance, thus highlighting its suitability for addressing the research objectives, albeit not at
an optimal level.
6. CONCLUSION
Contemporary research employs deep learning in image processing and pattern recognition,
facilitated by the intricacy of its designed neural architectures. These approaches have been
proven effective in identifying diseases that affect plant leaves. We revealed by our observations
that splitting and subsequently merging different leaves afflicted by distinct diseases results in a
complex structure suitable for training through convolutional networks. Augmenting the training
dataset also enhances the quantity of data available for training, leading to improved model
performance. Increasing the number of epochs impacts model accuracy. Consequently, an
efficient model capable of swiftly detecting multiple diseases on a plant leaf gets modeled using
this concept, demonstrating a confident prediction accuracy of 99.82% for two diseases.
In future research, with our work's objective in mind, we propose creating a more robust model
for tomato disease prediction. This model should be capable of predicting the number of diseases
present on an image while assessing their severity in terms of how much affected the leaves are
using the data fusion method. Additionally, constructing a more robust convolutional neural
network that can comprehend complex structures such as Disease fusion would be valuable for
future research, which is pertinent as most studies often involve only a single disease per leaf
when modeling with CNNs. The employment of a separable convolutional network addition of
13. International Journal on Cybernetics & Informatics (IJCI) Vol.12, No.7, December 2023
43
dropouts with batch transformation could be included in the model to enhance the training phase
with more profound learning abilities.
REFERENCES
[1] Shijie, Jia, Jia Peiyi, and Hu Siping. "Automatic detection of tomato diseases and pests based on leaf
images." 2017 Chinese automation congress (CAC). IEEE, 2017.
[2] Chen, Hui-Ling, et al. "Support vector machine based diagnostic system for breast cancer using
swarm intelligence." Journal of medical systems 36 (2012): 2505-2519.
[3] Blancard, Dominique. Tomato diseases: identification, biology and control: a colour handbook.CRC
Press, 2012.
[4] Arnal Barbedo, Jayme Garcia. "Digital image processing techniques for detecting, quantifying and
classifying plant diseases." SpringerPlus 2.1 (2013): 1-12.
[5] A. M. Mutka and R. S. Bart, "Image-Based Phenotyping of Plant Disease Symptoms," Frontiers in
Plant Science, vol. 5, pp. 1-8, 2015.
[6] S. P. Mohanty, D. Hughes and M. Salathe, "Using Deep Learning for Image-Based Plant Disease
Detection," eprint arXiv:1604.03169, 2016.
[7] Kaggle.com,https://www.kaggle.com/datasets/arjuntejaswi/plant-village?select=PlantVillage,
accessed on September 2023.
[8] Hong, Huiqun, Jinfa Lin, and Fenghua Huang. "Tomato disease detection and classification by deep
learning." 2020 International Conference on Big Data, Artificial Intelligence and Internet of Things
Engineering (ICBAIE). IEEE, 2020.
[9] Tm, Prajwala, et al. "Tomato leaf disease detection using convolutional neural networks." 2018
eleventh international conference on contemporary computing (IC3). IEEE, 2018.
[10] Alex Krizhevsky, Ilya Sutskever, and Geoffrey E Hinton. “Imagenet classification with deep
convolutional neural networks”. In: Advances in neural information processing systems.2012, pp.
1097–1105.
[11] Szegedy, Christian, et al. "Going deeper with convolutions." Proceedings of the IEEE conference
on computer vision and pattern recognition. 2015.
[12] Yann LeCun et al. “Backpropagation applied to handwritten zip code recognition”. In: Neural
computation 1.4 (1989),
[13] Usama Mokhtar et al. “Tomato leaves diseases detection approach based on support vector
machines”. In: Computer Engineering Conference (ICENCO), 2015 11th International. IEEE. 2015,
pp. 246–250.
[14] Ashqar, Belal AM, and Samy S. Abu-Naser. "Image-Based Tomato Leaves Diseases Detection
Using Deep Learning." IJARWpublication(2018).
[15] Llorca, Charmaine, May Elsbeth Yares, and Christian Maderazo. "Image-based pest and disease
recognition of tomato plants using a convolutional neural network." Proceedings of international
conference technological challenges for better world. 2018.
[16] Islam MP, Hatou K, Aihara T, Seno S, Kirino S, Okamoto S. “Performance prediction of tomato
leaf disease by a series of parallel convolutional neural networks.” Smart Agricultural Technology,
2022; 100054: 25 pp.
[17] Showmick Guha Paula, Al Amin Biswas, Arpa Saha, Md. Sabab Zulfiker, Nadia Afrin Ritu. “A
real-time application-based convolutional neural network approach for tomato leaf disease
classification.” Array19(2023)100313, 2023, 14 pp
[18] Geetharamani G. a, Arun Pandian J.“Identification of plant leaf diseases using nine-layer deep
convolutional neural network.” Computers and Electrical Engineering 76 (2019) 323-338.
https://doi.org/10.1016/j.compeleceng.2019.04.011
[19] Zhang Y-D, Dong Z, Chen X, Jia W, Du S, Muhammad K, et al. “Image based fruit category
classification by 13-layer deep convolutional neural network and data augmentation.” Multimed
Tools Appl 2019;78(3):3613-32.