Shahin Alom, ID Number of the Department of Computer Science and Engineering, Daffodil International University, has been accepted as satisfactory in partial fulfillment of the requirements for the degree of BSC in Computer Science and Engineering and approved in style and content. Department of Computer Science and Engineering Faculty of Science and Information Technology Daffodil International University. We therefore confirm that we have participated in this initiative at Daffodil International University under the leadership of Shayla Sharmin, Senior Lecturer in the Department of Computer Science and Engineering.
Our decision to pursue this project was inspired by our supervisor's tremendous insight and genuine intrigue in the subject of machine learning. Touhid Bhuiyan, Head of the Department of CSE, as well as to the other faculty member and employees of the CSE Department at Daffodil International University, for their kind consultation in concluding our research. We would like to thank everyone who was involved in this deliberation while doing the courses at Daffodil International University.
This study report consists of a pilot for the identification and classification of Paddy disease using inflamed leaves and algorithmic learning devices. The careful pre-handling of the photograph is the most crucial aspect of this study.
INTRODUCTION
- Introduction
- Motivation
- Rational of the Study
- Research Questions
- Expected Outcomes
- Project Management and Finance
- Report Layout
- Preliminaries
Despite the fact that most of our rural population lacks even basic knowledge of education, technology provides them with enough information. The treatment and possible resistance of the disease can be known after the diagnosis of the disease. Another intriguing aspect of the idea is that farmers can contact agricultural listings at any time with questions by phone or via the convenient chat feature.
To make our Android app more user-friendly, English and Bangla should be at least two of the main dialects. In an attempt to get a detailed overview of our research on “Detection of Disease on Paddy Leaves using Learnings tool” in Chapter 1. We can focus on the related works, the quick review of the synopsis, the aspects covered in our research are used, the breadth of the issues and the challenging circumstances of conducting this research.
We will briefly discuss the tools we used, the methods by which we collected and used our data, a statistical analysis of the data we obtained, the types of policies we used, and the requirements for implementing our indicated variant. Reflecting our activities on society and the environment may be the most important talking point in chapter 5.
BACKGROUND
- Related Works
- Comparative Analysis and Summary
- Scope of the Problem
- Challenges
One thousand images of diseased rice leaves were used to determine the accuracy of the Bayesian and SVM classifiers. Paddy Plant Leaf: Automated Blast Maldy Detection and a Color-Reducing Mechanism Maninder Lal Singh's method is detailed in the publication [4], which was generated with the help of Amandeep Singh. A unique mechanism for the propagation of paddy crop blast disease has been proposed in this paper.
It was found that the proposed calculation accuracy increased by 96.6% compared to comparisons with data from [15] and [16]. These techniques can be used in live initiatives in real-time histogram-based tactics that have been evaluated in and for color detection after checking circumstances that occurred in different environmental time frames. Based on this observation, a framework was developed to investigate rice bacterial blight, rice effect, brown spot disease and rice sheathing.
Classifiers, adequate-NN and MDC were used in the proposed approaches for the four mentioned rice diseases, and in my opinion they performed with an accuracy of 87.02% to 89.23%. Since all our statistics are raw data, we did not collect any images from the Internet.
RESEARCH METHODOLOGY
Research Subject and Instrumentation
Data Collection Procedure
Statistical Analysis
- Sample Image
- Magnaporthe Grisia (Blast Disease)
- Cnaphalocrocis medinalis (Rice Leafroller Disease)
- Fulgoromorpha (Plant Hopper Disease)
Magnaporthe grisea is a fungus that is dangerous to plants and promotes a chronic case in rice. It is also known by the names rice rot neck, rice blister fungus, rice seedling rust, rice blister, oval leaf spot of graminea, pit disease, ryegrass blast and Johnson spot. Magnaporthe oryzae was given to the very last members of the complex isolated from rice and countless other harbors.
Both names are now used by different writers, therefore it is unclear which to use for the rice blast pathogen. Members of the Magnaporthe grisea cluster can transmit diseases such as blast or blight disease to other essential agricultural grains, especially wheat, rye, barley and pearl millet. It is expected that enough rice will be broken each year to feed more than 60 million people.
On all parts of the shoot, the first symptoms appear as white to gray-green lesions or spots with darker borders, while older lesions are elliptical or fusiform, whitish to gray and have necrotic margins. The culm breaks off at the inflamed node due to a node infection (rotten neck). The rice leaf roller, or Cnaphalocrocis medinalis, is a genus of moths belonging to the family Crambidae.
It is established in Southeast Asia along with Sri Lanka, Taiwan, Thailand, Hong Kong and large parts of Australia. The larvae of the species Zea mays, Oryza sativa, Triticum, Saccharum and Sorghum are considered a pest. On the underside of the leaves, which may be scaly white in color, eggs are placed singly or in clusters and are axially aligned.
When attacking younger seedlings, it folds three to four adjacent shoot leaves and grinds away the green tissue, turning the infected leaves white. Any insect with more than 12,500 identified species worldwide and belonging to the suborder Auchenorrhyncha of the suborder Fulgoromorpha is called a planthopper. All members of this association, which is spread throughout the planet, are herbivores, but only a small number are considered pests.
Applied Mechanism
- CNN Model Configuration
- CNN Model
- Architecture of Our Model
- Transfer Learning Models
- Importing VGG16
- Importing Densenet201
- Importing Xception
Since we used four switches to get to know models (VGG19, VGG16, Resnet50, Xception), we can have a quick dialogue about them in the subsequent element.
Implementation Requirements
- Image Resizing
- Normalization
- Splitting of the dataset
- Featuring Data and labeling the class
We were then instructed to evaluate how effectively the tool retrieved from the data set.
Experimental Setup
- VGG16
- Architecture of VGG16
- VGG16 Estimator
- Performance of VGG16
- Visualization Graph of Accuracy and Loss
- Output of VGG16
- DenseNet201
- Architecture of DenseNet201
- Input pipeline of DenseNet201
- Performance of DenseNet201
- Xception
- Architecture of Xception
- Summary of Xception
- Performance of Xception
- Visualization of Xception
- Output of Xception
The DenseNet-201 architecture consists of a set of dense blocks, where each dense block contains a set of convolutional layers that are densely connected to the previous layers. In a dense block, each layer receives as input the feature maps of all previous layers, allowing the network to learn more efficient representations of the input. The output of the dense blocks is then passed through a series of transition layers that reduce the resolution of the feature maps and increase the number of channels.
The input to the network is an image, which is passed through dense blocks and transition layers to. The extracted features are then passed through a global averaging pooling layer and a fully connected layer, which perform feature classification. DenseNet-201 has achieved strong performance on a variety of image classification benchmarks, including the ImageNet dataset.
The input pipeline for DenseNet-201 typically involves preprocessing the input images before feeding them into the network. After the images are preprocessed, they are passed through the network in the form of a tensor, where they are convolved and down-sampled through a series of dense blocks and transition layers. The tensor is then passed through a global average pooling layer, which reduces the resolution of the feature maps, and through a fully connected layer, which performs classification on the extracted features.
The final output of the network is a probability distribution over the classes, indicating the probability that the input image belongs to each class. In addition to the main classification branch of the network, some architectures also include a localization branch, which is trained to predict the coordinates of the bounding box of an object in the input image. The localization branch consists of additional convolutional and fully connected layers that are added to the main classification branch of the network.
DenseNet-201 is a convolutional neural network architecture that has been widely used for image classification and segmentation tasks. DenseNet-201 has also been used as a base model for a number of more recent image segmentation models. In addition to its strong performance in image classification tasks, DenseNet-201 has also been used for other computer vision tasks, such as object detection and face recognition.
In these tasks, DenseNet-201 has also performed well and has been widely adopted by researchers and practitioners. Overall, DenseNet-201 has shown high performance in various image classification and computer vision tasks and has become a widely used model in this field.
Experimental Results & Analysis
Discussion
IMPACT on SOCIETY, ENVIRONMENT and SUSTAINABILITY
- Impact on Society
- Impact on Environment
- Ethical Aspects
- Sustainability Plan
Making a nearby host on our very own device and participating in the utility of the server is our ultimate goal. We intend to work on these in the future to improve the utility's user experience.
SUMMARY, CONCLUSION, RECOMMENDATION AND IMPLICATION FOR FUTURE RESEARCH
Summary of the Study
Conclusions
Implementation for Further Study
Plagiarism Report
