Journal of Computing & Biomedical Informatics Volume 04 Issue 02

ISSN: 2710 - 1606 2023

Research Article
https://doi.org/10.56979/402/2023

Early Detection and Classification of Rice Brown Spot and Bacterial Blight
Diseases Using Digital Image Processing
Junaid Iqbal1, Israr Hussain1*, Ayesha Hakim1, Sami Ullah2, Hafiz Muhammad Yousuf1

Department of Computer Science, MNS-University of Agriculture, Multan, Pakistan.

1

2Department of Agribusiness and Applied Economics, MNS-University of Agriculture, Multan, Pakistan.

* Corresponding Author: Israr Hussain. Email: israr.hussain@mnsuam.com.

Received: October 29, 2022 Accepted: February 23, 2023 Published: March 29, 2023.

________________________________________________________________________________________________________

Abstract: Rice (Oryza sativa) is an important food and a wonderful source of nourishment. It is
important to Pakistan's economic development. It is ranked as Pakistan's second-most significant
crop production region, behind wheat, with an estimated 2.31 million hectares. Each year, various
diseases in Pakistan cause the loss of 40% of the rice harvest. Pakistan has received reports of
numerous rice diseases, including Brown Spot and Bacterial Leaf Blight. In the farming area, rice
illnesses have reduced crop productivity and cost money. The exploration system aims to identify
disease symptoms on rice leaves. These disorders are first classified using image processing after
identification. We take several pictures of both healthy and diseased leaves. Additionally,
characteristics are extracted following image preprocessing. Images of the rice leaves are classified
as either healthy or diseased. When infected, it recognizes and categorizes properly. The model's
Inception v3 and VGG19 were used as classifiers. VGG19 outperformed with 97.94% accuracy.

Keywords: Rice Leaf Brown Spot, Bacterial Blight, Convolutional Neural Networks, Deep Learning,
Transfer Learning, Image processing, VGG19.

1. Introduction
The introduction of rice (Oryza sativa) is one of the most significant food, and half of the world pop-
ulation and contributing significantly to Pakistan's economy. Rice represents for 2.7 percentage of agricul-
tural fee delivered and 0.7 percent of GDP[1]. Pakistan's economy and yield greatly influence the produc-
tion of rice harvests. There are two prevalent diseases: brown spot (Figure 1) and bacterial leaf blight (Fig-
ure 2). Both the first and second diseases, which can damage rice health and reduce yields by up to 70 to
80 percent, are fungal and bacterial, respectively [2]. 40 percent of Pakistan's rice crop losses were reported
[3]. In Asia, bacterial leaf blight is the oldest rice disease. In order to prevent losses in yield production, it
is important to identify rice diseases. It takes a lot of time and effort to recognise and identify rice leaf
diseases [4]. Due to their similarities, classifying the rice leaf diseases is challenging. It is difficult to identify
and categorise them by human perception at an early stage. It would be more appropriate and beneficial
to use the image processing approach for the early identification and categorization of rice illnesses. Using
digital image processing and machine learning, the suggested approach correctly detects diseases in the
early stages of the rice crop.
Brown spots (Alternaria longipes) are fungi that harm spikelets, leaves, rice husks, coleoptiles, and
leaf sheaths. The most noticeable harm is the leaf's abundance of big spots, which can kill the leaf as a
whole. Unfilled grains, blotches, or discolored seeds develop when seeds get sick. One of the most wide-
spread and destructive rice illnesses in recorded history, brown spot disease, has received little attention.

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02
Loss of volume and quality are both caused by brown patches[5]. Brown spot (Figure 1) in the field is
mostly caused by, infected seed, which gives rice to infected seedlings[4].

Figure 1. Brown Spot Leaf

One of the deadliest bacterial infectious illnesses to ever affect farmed rice is rice blight, often known
as rice blight (Oryza sativa and O. glaberrima) [6]. A serious epidemic that affects millions of hectares of
rice each year can result in crop losses of up to 75%. The disease was first observed in Kyushu, Japan from
1884 to 1985. Breeding mortality in warm and humid environments has been observed in rice-growing
regions of Asia, the west coast of Africa, Australia, and the Caribbean Sea. Although not commonly found
in the United States, strains of bacteria associated with Xoo are listed as agricultural selection agents by the
United States Department of Agriculture. It is a designation that is subject to strict regulation and Bacterial
Blight disease as shown in (Figure 2).

Figure 2. Bacterial Blight Leaf

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02

The main contribution to this study is;

1. To find optimal texture features to achieve high accuracy for the detection of Brown Spots and Bacterial
Blight.
2. To develop an intelligent system for the early detection and classification of rice diseases.
3. To design an online platform for the visualization of experimental results.

2. Literature Review
The Materials The plant leaf disease system proposed by [7] was developed for detection and classifi-
cation using computer vision and machine learning techniques. Experimental results were evaluated by
Random Forest, Support Vector Machines, K Neighbor Neighbors, and Artificial Neural Networks and
compared with various algorithms. The performance of the proposed predictive model was better in Ran-
dom Forest compared to other classifiers. This is to provide 74.28% accuracy with a 70.89% F-score, 71.88%
recall, and 71.77% precision score.
An automated system was developed by [8] to detect rice leaf disease using image-processing tech-
niques. Automatic Plant Disease Leaf Detection Various hybrid image segmentation and classification
techniques have been recognized and analyzed to identify virtual disease of rice leaves in different envi-
ronments. Characteristic-based detection of rice leaf disease using a support vector machine and multiple
convolutional neural networks (CNNs). The deep features proposed by the system are categorized by Sup-
port Vector Machines (SVMs). The transfer learning approach was applied to detect rice leaf disease using
a deep convolutional neural network. The highest classification accuracy of the system was 97.31% using
features and the lowest classification accuracy was 94.40% using features [9].
Identification of rice diseases and classification by color characteristics. The proposed system designed
was an automated system for the detection and classification of rice diseases. Color features were extracted
from rice diseases. Several classifier algorithms were compared and the highest accuracy of 94.68% was
achieved using a Support Vector Machine (SVM) classifier [10].
Methodologies for detecting and classifying rice diseases include image preprocessing, segmentation,
and feature extraction. The proposed system was classified using both the k-nearest neighbor method (k-
NN) and the minimum distance classifier (MDC) and evaluated images of rice leaves of various diseases.
In addition, 70% of the image data was used for training and the other data was used for testing. The
accuracy results were 87.02% for K-NN and 89.23% for MDC [11].
The proposed system uses image processing to identify and classify paddy disease diseases based on
the RGB values of the infected area. This system has succeeded in identifying several rice diseases such as
rice blast, rice blast, and rice blast. This system produces 90% classification accuracy for brown spots, 90%
for bacterial leaf bright, and 89% for rice blasts [12].
Many fungal and bacterial diseases infect rice leaves and nodes at various stages, causing loss of
growth and production. The proposed system was developed by [13] as an automated system for identi-
fying and classifying many diseases of rice. This method detects some diseases in a wide area of rice culti-
vation. The system also improves accuracy performance.
The suggested solution uses machine learning and image processing to increase the accuracy of iden-
tifying illnesses in rice leaves. The proposed approach made use of numerous classification strategies for
rice leaf diseases. Results of rice leaf diseases are classified and provided with an accuracy of 94.6 percent
using the Histogram of Oriented Gradients (HOG) and Support Vector Machine [14].
The project aims to develop a software system that can automatically identify and categorize diseases.
Utilizing the Minimum Distance Classifier (MDC) and the k-Nearest Neighbor classifier, all of the collected
features were combined according to the diseases (k-NN). The classification accuracy for each disease was
calculated using both classifiers. The overall accuracy while using k-NN and MDC is 87.02 percent and
89.23 percent, respectively. For two diseases, rice blast, and rice brown spot, both classifiers have the same
classification accuracy; however, for the other two diseases, blight and sheath rot, MDC has a greater clas-
sification accuracy than the k-NN classifier [15].
Early diagnosis of rice illnesses, particularly those affecting the leaves, enables farmers to take the
required safeguards early on for higher-quality crops. It is usually desirable to use RGB color photographs
as inputs and to offer analysis results even with RGB color images when creating an automated system for

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02
classifying and recognizing rice blight. The study suggested a suitable framework that combines improve-
ments, filters, color segmentation, and color functionality for identification classification processes. To im-
prove the detected accuracy rate, a CNN classifier was used. This framework, suggested by the poll, offers
an acceptable accuracy rate of 91.43 percent in comparison to other existing methodologies[16].
Rice output can be boosted by identifying rice pests and illnesses and effectively managing paddy
areas where pests are common. Farmers can help other farmers recognize and identify the many kinds of
pests and diseases in their rice fields with the aid of cutting-edge technology, such as cell phones. Find rice
diseases by utilizing a convolutional neural network and the r programming language to analyze photos
from the disease sheet. Three diseases—bacterial leaf blight, brown spots, and leaf smut are among the
disease patterns gathered from the UCI Machine Learning Repository. The training image can be improved
to produce better outcomes[17].
One of the intriguing research areas in computing and agriculture is the detection of illnesses using
photographs of plants. To diagnose rice diseases from photos of damaged rice, the study paper presents
an overview of several image processing and machine learning techniques. In addition to outlining the
various methods, the white paper also provides a succinct overview of the main image processing and
machine learning ideas that are relevant to the identification and classification of plant diseases. 19 articles
on various plants and fruits, including rice illnesses, will be examined in more detail, and they will be
summarised based on important factors. The size of the image dataset is one of these factors. categoriza-
tion of the class (illness), preprocessing, segmentation technique, classifier kind, classifier precision, etc.
We propose and shape work on the characterization and classification of illnesses in rice plants using re-
search and more study[18].
In the agricultural sector, rice diseases have considerably decreased output and cost money. To man-
age and lessen the effects of assaults, early disease detection is crucial. Early detection of disease severity
and outbreaks can help to prevent production from suffering both quantitative and qualitative losses while
also reducing the demand for pesticides and fostering national economic growth. This study describes an
integrated approach for diagnosing Rice Blast, a leaf disease, using image processing techniques (RLB).
The image preprocessing, image segmentation, and image analysis processes all employ the Hue Satura-
tion Value (HSV) color space. Areas of interest are extracted using the most important image processing
technique, picture segmentation, and a multi-level thresholding approach-based pattern recognition
method is offered. As a result, the infection stage, the dispersion stage, and the worst stage of the RLB
disease may be identified. This technique successfully detects disease in images taken in an uncontrolled
environment [19].
By developing disease-resistant varieties, rice bacterial leaf blight (BLB) can still be economically and
efficiently controlled. Breeders have contributed greatly to the development of BLB-resistant rice varieties.
The dangerous illness known as BLB infection has been found to significantly reduce rice production. Pick-
ing rice with many resistance genes will be challenging using just the traditional method. This is due to the
masking effect that genes, including epistasis, have. Furthermore, using traditional breeding techniques, it
takes a very long period to develop the target gene. Interconnect resistance is a big challenge with tradi-
tional approaches. Through the use of markers in molecular breeding, the identification and introduction
of BLB disease-resistance genes were made simpler. Rice genetic engineering (introduction) is another bi-
otechnology-enabled invention. Although a molecular approach to combating BLB disease in rice farming
is beneficial, the increasing demand for rice farming in a rapidly growing societal population prevents
molecular breeding from replacing traditional breeding. The research focuses on recent advances in creat-
ing BLB disease-resistant rice cultivars utilizing molecular methods rather than conventional methods [20].
Accurate and early diagnosis of rice diseases and pests enables farmers to process rice quickly and
significantly lower economic loss. Deep learning-based convolutional neural networks (CNNs) have lately
made significant progress that has significantly improved the accuracy of picture classification. Based on
CNN's success in picture categorization, this article has built a deep learning-based system for identifying
diseases and pests from photographs of rice. There are two contributions in this publication. (I) Inception
V3 and VGG16, two contemporary large-scale architectures, have been employed and enhanced for this
purpose. Experimental results show how effective these models are using datasets from the real world.
Because current CNN architectures like MobileNet, NasNet Mobile, and SqueezeNet are not appropriate
for mobile devices, a two-tiered small CNN design was proposed and contrasted. Experimental results

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02
show that the proposed design may significantly reduce model size while maintaining the required accu-
racy of 92.3%[21].

3. Proposed Model
The automatic diagnosis of rice diseases using the convolutional neural networks (CNNs) model was
developed in the research study by [22]. Images of healthy and sick rice crops were taken using datasets of
those crops. The CNN algorithm was trained to recognize various widespread illnesses affecting rice. Most
rice-variety diseases are universal. Using image processing technology for the early identification and clas-
sification of rice diseases would be more useful and relevant in the proposed methodology (Figure 3)

Images Dataset Selection

Images Dataset Preprocessing

Build Classifier Model

Compile Classifier Model

Model Training and Validation

Model Evaluation

Classification

Results

Figure 3. The Basic Steps of Methodology

Using digital image processing and machine learning, the suggested approach correctly detects dis-
eases in the early stages of the rice crop. The goal of its research is to create a prediction and classification
model that can safeguard farmers from dangers associated with rice production. Farmers detect appropri-
ate characteristic combinations, such as temperature, water availability, and weather, to determine which
performs best and how much rice yield production can be expected based on these values. Those features
determined by machine learning will be more important and have a favorable effect on rice crop growth.
3.1. Material and Methods
This chapter briefly describes the methodology of the proposed system. The proposed methodology
(Figure 4) contains six phases.
1. Image Acquisition
2. Image Pre-processing
3. Feature Extraction
4. Optimized Dataset
5. Classification Model
6. Visualization
3.2 Proposed Methodology Workflow

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02

Figure 4: Methodology Framework

3.2.1 Image Acquisition

A dataset collection of several raw images of rice leaves makes up the first stage. We call for healthy
leaf photos as well as photographs of illnesses like brown spots and bacterial leaf blight from various ex-
isting datasets. The gathered photographs are kept in JPEG formats. Rice disease datasets are collected for
their detection and classification. The rice disease dataset contains 480 images of the two most popular
commercial diseases, Brown Spot and Bacterial Blight (more than 140 images of each variety). To identify
and classify the above diseases, the dataset was divided into two categories.
High-resolution digital photography and recording devices are more available and cheaper. Addi-
tionally, SEM and associated image acquisition methods offer a way to capture images with a resolution
that contains both structural and elemental data. Choosing the best imaging option for your research chal-
lenge is still a crucial step in sediment research, despite the emergence of novel acquisition methods and
gear. In addition, there are common methodologies and concerns for acquiring the best images for quanti-
tative sediment analysis despite the diverse acquisition techniques. To quantify differences in characteris-
tics between photographs, each image must also contain color or an equal image standard. Last but not
least, your research objectives should be appropriate for the image resolution and file storage procedure,
which should preserve the original image information[15].

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02

Figure 5. Brown Spot, Bacterial Blight, and Healthy Leaf

3.2.2 Image Pre-processing

By showcasing the best outcomes of the image, image enhancement is a key goal of image prepro-
cessing. These procedures, which involve resizing, enhancing the image quality, and color correction, are
carried out before the model training [23]. The detection of local and global features can be improved by
making intelligent use of picture preprocessing, which can also solve issues. Preprocessing is done to en-
hance the image's quality so that it may be better evaluated. Pretreatment can reduce unwanted distortion
and improve some of the characteristics required for the particular application you are working with. These
features may vary from application to application. Image processing is crucial to computer vision.
The process of turning an image into a digital representation and then executing operations on it to
extract important information is known as image processing. Image preprocessing is an integral part of the
classification of images processed for research purposes. The need for pretreatment arises from a variety
of lighting conditions, dimensional differences, and so on[24]. Input image preprocessing is typically done
on the CNN`s input layer. Many pre-processing such as normalization, trimming, resizing, etc. may be
required. However, CNN required much less preprocessing. Next, is another neural network. However,
simple pre-processing is required to remove unwanted parts of the image. The ratios for the training and
test datasets are 80% and 20%, respectively. The CNN classifier (test V3) was trained using a training da-
taset.
3.2.3 Feature Extraction
The best outcome of the provided dataset is used for the feature extraction. The texture feature will be
used to extract details from various pixels' brightness and intensity. The characteristics can be applied to
rice leaves' diseased parts [25]. The two fundamental building blocks of all pattern recognition and com-
puter vision systems are feature extraction and classifier designs. The most challenging but crucial phase
in the analysis of visual patterns is the extraction of robust and discriminating characteristics from an im-
age. Several common and sophisticated methods for extracting features from photos are being taken into
consideration, some of which have already undergone in-depth analysis. Here is an illustration of how to
use these feature extraction techniques to address challenges with image-based disease detection and clas-
sification. These use examples illustrate concepts, theories, and innovations in feature extraction methods,
as well as their benefits and drawbacks in dealing with practical issues [26].
3.2.4 Optimized Dataset
Following feature extraction, these characteristics will be refined to create an optimal dataset that
takes advantage of and analyses data from each class in real time. Consequently, the feature optimization
approach might be useful for the problem [27] by selecting only pertinent data and eliminating noise from
the data, feature selection is a technique for lowering the number of model input variables. This approach
involves choosing features for your machine learning model on an automatic basis based on the kind of
problem you are attempting to address. Utilize various techniques to identify the most useful dataset [28].
3.2.5 Classification Model
Machine learning classifiers come in a variety of forms. Our dataset will determine which categoriza-
tion model we use [29]. A structural neural network with many layers is called a convolutional neural

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02
network. The convolutional layer is the top layer of the CNN network. The majority of the computing
power is handled by this fundamental building piece. A filter or kernel is used to reduce the size of the
data or image. A sliding window allows you to apply a small unit called a filter to your data. Others ML
classifiers like KNN, Inception V3, and, VGG19 were also deployed.
3.2.6 VGG19
A 19-layer, the collapsible neural network is called VGG19. From the ImageNet database, you can
load a network that has already been trained using more than 1 million images. You may classify your
photographs using a pre-trained network into 1000 different object categories, including keyboards, mice,
pens, and various animals. The network has therefore acquired a comprehensive feature representation of
a variety of images. The network image's input dimensions are 250x250. Over the VGG19 network, classi-
fication enables you to classify fresh images. Observe the procedures for using Google Neural Network to
classify photographs, then use VGG19 in its place. Follow the training instructions for the deep learning
network to identify the new photos and load VGG19 (Figure 6) rather than GoogleLeNet to retrain the
network for a new classification assignment [29].

Figure 6. Image Detection Using VGG-19 with CNN

4. Results
The proposed work's experiments have produced positive classification results, with the maximum
average classification accuracy using the first KNN model accuracy being 67.18%, and second Inception
V3 model at 93.57% and the final VGG-19 model being 97.94% respectively. The classification results show
that the suggested Inception V3 and VGG-19 CNN models performed nearly as well as the outcomes re-
ported in recently published works. The proposed method has greater potential for classifying many forms
of rice diseases with more variety of symptoms than prior studies carried out by other researchers, there-
fore the overall classification performance is satisfactory. The parameters that we used in our work are
given in Table 1.
Table 1. Values of the parameters of VGG19 used in our experimental

Parameters Parameters Values

No. of Layers 24

Learning Rate 0.001

Dropout 0,5

Validation Threshold 20

Epoch 10(min), 20(max)

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02

Training Accuracy

100

80

60

40

20

0
VGG19 InceptionV3 KNN

Accuracy

Figure 7. Comparison of Training Accuracy

Figure 7 and Figure 8 show the training accuracy and validation accuracy results of the models that are
used in this study. VGG19 was more suitable than others as the results show by getting high accuracy.
Overall results comparison are shown in figure Figure 9 as VGG19 topping the chart.

Validation Accuracy

100

80

60

40

20

0
VGG19 InceptionV3 KNN

Validation Accuracy

Figure 8. Comparison of Validation Accuracy

5. Discussion
The comparison outcomes of the suggested methodology to cutting-edge publications. The literature
reports work on the use of various image processing and machine-learning techniques to identify diseases
affecting rice leaves. The classification of rice disease symptoms, however, is done in the current work
employing cutting-edge deep learning models using fresh datasets and combinations. The majority of the
literature is concentrated on a small number of disease types identified on rice leaves. Compared to an
earlier study by other researchers, the dataset includes a wider variety of rice diseases with a wider range

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02
of symptoms. The suggested approach takes into account the automatic detection of rice diseases brown
Spot and Bacterial Blight that affect rice leaves. Although baseline training, also known as training from
scratch, performs better than transfer learning, the results are still unsatisfactory. Less than 85% of the two
models are accurate. To discover sets of the kernel to extract good discriminative features, CNN uses a set
of kernel matrices as a local feature extractor to extract regional features. The hyperparameter needed to
be adjusted in the proposed model. Other techniques, such as K-NN and Inception V3 classifiers, are also
employed for classification. But third and VGG19 model accuracy is above 95%. By gathering data on rice
leaves, this model also aids in disease identification. Performance outcomes of VGG19-based deep feature
extraction and transfer learning. Low precision is also indicated by the enormous standard deviation of the
validation accuracy.

97.9496.69
100 93.57
90.43
90
80
67.23 65.9
70
60
Accuracy
50
VAccuracy
40
30
20
10
0
VGG19 InceptionV3 KNN

Figure 9. Comparison Analysis

6. Conclusion
The tests conducted on the database of rice leaf diseases, including Brown Spot and Bacterial Blight,
are analyzed in this work. Many pictures of both healthy and diseased leaves have accumulated. Images
of the leaves will be used to classify them as either healthy or infected. The proposed system prediction
and classification of diseases of rice leaves. This is beneficial not only for farmers but also for exporters.
Rice leaf disease predictions are rice estimates using received historical data such as weather and rice crop
attributes. It not only provides information on rice leaf disease but also estimates important parameter
values for rice that farmers need to take into account for maximum production. We will also explain the
results of the experiment. In the comparative analysis with the various models performed in this paper,
the results are obtained and explained using the proposed methodology using graphs and tables. Results
are obtained at 80% and 20%, respectively, depending on the dataset distributions of the training and test
datasets. The implementation of the proposed methodology is carried out for the best results. Additionally,
we will use prototype picture acquisition and machine vision models for the early detection, categoriza-
tion, and determination of intelligent real-time rice cultivation. The implementation of the proposed meth-
odology is carried out for the best results.KNN accuracy is 67.18%, and Inception V3's automated system
produces 93.57% accuracy results. And the other best automation system with VGG19 accuracy results is
97.94%. In general, in deep learning of dataset distributions, the larger the training dataset and the higher
the quality of the dataset, the better the performance. To increase Pakistan's rice exports, the proposed
methodology can be put into practice on a wide scale for automatic identification and categorization by
global standards.

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02
References:
1. N. Shahzadi, M. Akhter, Z. Haider, U. Saleem, and A. Mahmood, "Rice in Pakistan: present scenario, trade, problems and pro-
spects," Int J Agric Stat Sci, vol. 14, pp. 1-6, 2018.
2. A. Rafi et al., "Field based assessment of rice bacterial leaf blight in major rice growing zones of Pakistan," Sarhad Journal of
Agriculture, vol. 29, no. 3, pp. 415-422, 2013.
3. M. H. Masood, H. Saim, M. Taj, and M. M. Awais, "Early disease diagnosis for rice crop," arXiv preprint arXiv:2004.04775, 2020.
4. S. Ramesh and D. Vydeki, "Rice blast disease detection and classification using machine learning algorithm," in 2018 2nd Inter-
national Conference on Micro-Electronics and Telecommunication Engineering (ICMETE), 2018: IEEE, pp. 255-259.
5. H. M. U. Aslam et al., "First report of brown leaf spot of rice caused by Bipolaris zeicola in Pakistan," Plant Disease, vol. 105, no.
1, p. 212, 2021.
6. J. Zhang, Y. Yang, X. Feng, H. Xu, J. Chen, and Y. He, "Identification of bacterial blight resistant rice seeds using terahertz imaging
and hyperspectral imaging combined with convolutional neural network," Frontiers in Plant Science, vol. 11, p. 821, 2020.
7. N. Ganatra and A. Patel, "A multiclass plant leaf disease detection using image processing and machine learning techniques,"
International Journal on Emerging Technologies, vol. 11, no. 2, pp. 1082-1086, 2020.
8. K. Archana and A. Sahayadhas, "Automatic rice leaf disease segmentation using image processing techniques," Int. J. Eng. Tech-
nol, vol. 7, no. 3.27, pp. 182-185, 2018.
9. P. K. Sethy, N. K. Barpanda, A. K. Rath, and S. K. Behera, "Deep feature based rice leaf disease identification using support vector
machine," Computers and Electronics in Agriculture, vol. 175, p. 105527, 2020.
10. V. K. Shrivastava and M. K. Pradhan, "Rice plant disease classification using color features: a machine learning paradigm," Jour-
nal of Plant Pathology, vol. 103, no. 1, pp. 17-26, 2021.
11. A. A. Joshi and B. Jadhav, "Monitoring and controlling rice diseases using Image processing techniques," in 2016 International
Conference on Computing, Analytics and Security Trends (CAST), 2016: IEEE, pp. 471-476.
12. T. Islam, M. Sah, S. Baral, and R. R. Choudhury, "A faster technique on rice disease detectionusing image processing of affected
area in agro-field," in 2018 Second International Conference on Inventive Communication and Computational Technologies
(ICICCT), 2018: IEEE, pp. 62-66.
13. R. Narmadha and G. Arulvadivu, "Detection and measurement of paddy leaf disease symptoms using image processing," in
2017 International Conference on Computer Communication and Informatics (ICCCI), 2017: IEEE, pp. 1-4.
14. M. E. Pothen and M. L. Pai, "Detection of rice leaf diseases using image processing," in 2020 Fourth International Conference on
Computing Methodologies and Communication (ICCMC), 2020: IEEE, pp. 424-430.
15. A. Devaraj, K. Rathan, S. Jaahnavi, and K. Indira, "Identification of plant disease using image processing technique," in 2019
International Conference on Communication and Signal Processing (ICCSP), 2019: IEEE, pp. 0749-0753.
16. T. S. Sazzad, A. Anwar, M. Hasan, and M. I. Hossain, "An Image Processing Framework To Identify Rice Blast," in 2020 Interna-
tional Congress on Human-Computer Interaction, Optimization and Robotic Applications (HORA), 2020: IEEE, pp. 1-5.
17. A. Sony, "Prediction of Rice Diseases Using Convolutional Neural Network (in Rstudio)," International Journal of Innovative
Science and Research Technology, vol. 4, no. 12, 2019.
18. J. P. Shah, H. B. Prajapati, and V. K. Dabhi, "A survey on detection and classification of rice plant diseases," in 2016 IEEE Inter-
national Conference on Current Trends in Advanced Computing (ICCTAC), 2016: IEEE, pp. 1-8.
19. M. A. Bakar, A. Abdullah, N. A. Rahim, H. Yazid, S. Misman, and M. Masnan, "Rice leaf blast disease detection using multi-level
colour image thresholding," Journal of Telecommunication, Electronic and Computer Engineering (JTEC), vol. 10, no. 1-15, pp.
1-6, 2018.
20. S. C. Chukwu et al., "Bacterial leaf blight resistance in rice: a review of conventional breeding to molecular approach," Mol Biol
Rep, vol. 46, no. 1, pp. 1519-1532, Feb 2019, doi: 10.1007/s11033-019-04584-2.
21. C. R. Rahman et al., "Identification and recognition of rice diseases and pests using convolutional neural networks," Biosystems
Engineering, vol. 194, pp. 112-120, 2020.
22. Y. Lu, S. Yi, N. Zeng, Y. Liu, and Y. Zhang, "Identification of rice diseases using deep convolutional neural networks," Neuro-
computing, vol. 267, pp. 378-384, 2017.
23. S. K. Upadhyay and A. Kumar, "A novel approach for rice plant diseases classification with deep convolutional neural network,"
International Journal of Information Technology, pp. 1-15, 2021.
24. G. Hemalatha and C. Sumathi, "Preprocessing techniques of facial image with Median and Gabor filters," in 2016 International
Conference on Information Communication and Embedded Systems (ICICES), 2016: IEEE, pp. 1-6.
25. R. Kumar, G. Baloch, A. B. B. Pankaj, and J. Bhatti, "Fungal blast disease detection in rice seed using machine learning," Interna-
tional Journal of Advanced Computer Science and Applications, vol. 12, pp. 248-258, 2021.
26. X. Jiang, "Feature extraction for image recognition and computer vision," in 2009 2nd IEEE international conference on computer
science and information technology, 2009: IEEE, pp. 1-15.
27. S. Ramesh and D. Vydeki, "Recognition and classification of paddy leaf diseases using Optimized Deep Neural network with
Jaya algorithm," Information processing in agriculture, vol. 7, no. 2, pp. 249-260, 2020.
28. [28] B. Biswal, P. K. Dash, and S. Mishra, "A hybrid ant colony optimization technique for power signal pattern classifica-
tion," Expert Systems with Applications, vol. 38, no. 5, pp. 6368-6375, 2011.

ID : 119-0402/2023
Journal of Computing & Biomedical Informatics Volume 04 Issue 02
29. M. A. Islam, N. R. Shuvo, M. Shamsojjaman, S. Hasan, S. Hossain, and T. Khatun, "An automated convolutional neural network
based approach for paddy leaf disease detection," Int. J. Adv. Comput. Sci. Appl, vol. 12, no. 1, pp. 280-288, 2021.

ID : 119-0402/2023