Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Artificial Neural Network Algorithms enhanced Chili

Leaves Diseases Identification system

ABSTRACT-Major developments in Deep learning (DL) and Machine learning (ML)


techniques have facilitated the recognition and detection of objects in images. Going to follow
their success in these other fields, neural nets have recently entered a variety of agricultural and
farming applications. Software for detecting plant diseases that can assist farmers in managing
their crops more efficiently, increasing yields. Using pictures to identify plant disease in crops is a
challenging challenge in and of itself. The application of specialized control techniques requires
both detection and individual species identification. A review of research projects using Google
Net, a form of DL, to address various plant disease detection issues was conducted for the
current publication. The dataset comprised 180 Chili diseased photographs deployed from the
Kaggle public domain and subjected to the Google-Net convolutional neural network (CNN) to
test the training accuracy. By considering various training properties such as CNN optimizers
SGDM, ADAM, and RMSPROP w.r.t Max Epochs and assigning Dropout probability, Strides,
Dilation factor, and padding values as constants, the obtained training accuracy ranges from 30%
to 100%. According to the simulation, the Google-Net CNN architecture achieves 98.61%
accuracy in SGDM and ADAM with Max Epochs of 20, 30, and 40, respectively. Moreover, it
was suggested that the applicability of SGDM is good for training the Chili Dataset, where Max
Epochs are much lower than in SGDM and ADAM. The three epochs in the two optimizers are
20, 30, and 40, with low epoch 20 generating higher accuracy.
Keywords: Image Processing, Deep learning, Chili leaf diseases, Machine learning, CNN,
Classification.

1. INTRODUCTION

The core agricultural products include chili. Due to the minimum quantity of chili, it
sometimes becomes appealing in the market. One of the most-risk plants is the chili plant.
Consequently, they adopt new technologies into a significant material to be measured due to
strategy and technological expertise. The purpose of growing structured and methodical must
be to increase the general generation by utilizing administrators. Many farmers refused to
cultivate chilies since the disease became a high threat to quality control and productivity to
generate chilies during the rainy season. There are two main types of factors that might result in
the devastation of chili plants, both for living and non-living entities. Bacteria, fungi, viruses,
and insects all contribute to compensating living beings.

Examples of non-living agents include temperature limits, inadequate lighting, nutrient


deficiencies, excessive moisture, low soil pH, and air pollution. Disease detection requires the
ability to recognize disease symptoms. Various things, such as irregular leaf development,
shriveled and damaged pods, stunted growth, and color distortion are examples of typical
symptoms. In order to comprehend the various ways to plant leaf disease detection, we will
look at a variety of writings by diverse authors in this essay. The section that follows will display
the works of numerous authors.
Sushil R. Kamlapurkar proposed a technique for categorizing and detecting diseases from
images of plants that can give more precise results. They had to use a range of procedures,
involving identification, training, and pre-processing. They had to conduct feature extraction,
classify the photos, and then perform the diagnosis. They identify diseases in this way [1].

Sachin D. Khirade et al. used image processing to find plant illnesses. The process entails
several processes, including as image pre-processing, segmentation, classification, feature
extraction, and picture collection. They discussed methods to identify the illness using picture
databases. [2]. Zulkifli Bin Husin and colleagues discuss the diagnosis of chili plant diseases
through the examination of leaf features. This system gathers and processes leaf photos in order
to assess the health of the plant. Image processing is employed to perform disease images.
They had to process the image in a variety of ways [3].

Meena Prakash as well as others R R. Meena Prakash et al. combine classification and
image analysis, two crucial techniques, to first detect the disease and then classify it. Support
vector machines are used by them for identification. The feature extraction process was carried
out using the Gray-Level-Cooccurrence Matrix. For segmentation, the k-means algorithm is
employed [4]. A technique to identify grape plant disease was put out by Pranjali B. Padol and
colleagues. They did this by using ANN and SVM classifiers as well as image processing
techniques. They unveiled a brand-new classifier that combines ANN and SVM. Regarding
grape leaf powdery and downy disease, they had 100% accuracy [5]. An image-processing
method was suggested by Baldomero Manuel Sanchez Rangel et al. for identifying and
classifying grapevine leaves as having a particular dimension of potassium deficit.

The K-Nearest Neighbors (KNN) segmentation technique was contrasted with histogram-
based techniques. When the environment in which images are acquired is less regulated, KNN
showed superior outcomes. [6]. Akanksha Rastogi and colleagues developed a straightforward
and computationally capable methodology for identifying plant disease and grading exploitation.
Digital image processing and machine vision This framework is split into two parts. The first
stage detects the concept of a leaf plant feature, which is then followed by pre-processing and
feature extraction. An Artificial Neural Network is used for classification. In the second phase,
the disease in the leaf is classified; for segmentation, the K-means algorithm was used, and for
feature extraction, an ANN-based feature extraction method was used. They used fuzzy logic
for gradation. [7].

Shivani K Tichkule and others Through image processing are able to identify numerous
plant illnesses. On plants, they found diseases brought on by fungi, viruses, or bacteria. Image
processing offers an effective method for illness diagnosis in images. K-means and ANN were
used for classification purposes. To keep tabs on agricultural activities, they had to utilize robots
[8]. The feature extraction process uses the Gabor filter and the Sobel edge detector. In this
paper, feature selection is used. The current MLP and how the suggested approach is used
is clarified. The leaf features, such as edges and textures, are extracted using MATLAB. These
visual attributes can be used to explain how pixels are arranged in an image.
According to the experimental results, the proposed NCSFFNN has 6.67% higher
classification accuracy than the MLP, and for real data, the proposed NCSFFNN has 8.72%
higher classification accuracy than the MLP and 1.92% higher classification accuracy than the
proposed NCSFFNN, respectively [9].

Boran Sekeroglu et al detected and recognized disease on 27 different types of leaves using
a back propagation neural network method, and their recognition rate is 97.2%. the use of
MATLAB, which includes the addition of noise to images as well as the changing and resizing
of image types in order to reduce processing costs [10]. This paper is arranged as follows:
Section II describes the Related works. Section III Presents the Methodology and Section IV
Presents the Experimentation. And Section V Describes the Results and Conclusions are drawn
in Section V.

2. RELATED WORKS

The survey takes into account 15 different types of AI-based disease detection case
studies. The table mostly displays AI-related ML and DL-based classification accuracies. Based
on the AI detection results from a review of the literature on 15 leaf diseases with varying illness
identification accuracy. The Deep Learning algorithms efficiently detect eleven ailments, while
the remaining four diseases are simply diagnosed using Machine Learning approaches.

SVM classifiers, which are machine learning algorithms, performed better than DL
approaches in detecting Anthracnose, Powdery Mildew, and infections. The remaining ML
approaches, including KNN (89.2 percent accuracy), Naive Bayes (89 percent), EKNN (87
percent), and PCA (98 percent), performed well in diagnosing diseases on the Plant Village,
CGLS, and Google datasets, with accuracy ranging from 63% to 99.5 percent. For each ailment,
the detailed accuracy findings were displayed in a table-1.

The performance of DL algorithms on leaf diseases detection, such as Black Spot,


Melrose, Corn Gray Leaf Spot, Cotton Leaf Diseases, Bacterial Blight, Alternaria, Cercospora,
Rice Blast Disease, Grey Mildew, Fusarium Wilt, and Stem Diseases, is pretty impressive and
results in high accuracy. Deep net models such as DCNN (96 percent), YOLO (96.63 percent),
CNN (99.48 percent), Exception, VGG-16 (97.89 percent), CNN-LSTM (92.55 percent),
Squeeze net (91 percent), and Resnet-18 achieved classification accuracies ranging from 92.6
percent to 99.4 percent. According to the table, Deep Learning algorithms outperform ML
techniques in most circumstances when it comes to identifying leaf illnesses.

Even though the ML algorithm is dominating in detecting leaf diseases, the difference in
accuracy is minor. However, because of the extensive set of convolution layers and strength
training sessions, the DL methods are effective in detecting leaf illness. Table 1 represents a
comparative analysis of ML and DL methods in classifying multi-leaf disease detection.
Table.1: Multi-Leaf disease detection classification accuracies based on ML and DL

Machine Learning Deep Learning


S.No Disease Name Dataset Accurac
Method Method Accuracy
y
1 Anthracnose Plant Village MSVM 95.6% BPNN 97..80%
2 Black spot Plant Village SVM 93.00% ANN 94.00%
3 Melrose Images DSLR VGG-16 89.5% D-CNN 96.00%
4 Corn Gray Leaf CGLS Random 79.23% conditional 99.79%
Spot subspace method random field
(CRF)
5 Powdery Mildew Plant Village SVM 98.65% MSO-ResNet 95.70%,
model
6 Cotton leaf Plant Village SVM 90.00% D-CNN 97.98%
Diseases
7 Bacterial Blight Google Decision tree 97.00% Optimized Deep 98.90%
algorithm Neural Network
with Jaya
Algorithm
8 Alternaria Plant Village Regression- 83.26% Gaussian Filtering 98.63%
Median Filtering

9 Cercospora Plant Village GLCM 94.04% SECNN 99.28%

10 Rice Blast Plant Village Q-SVM 81.8% CNN-AlexNet 93.79%


Disease
11 Grey Mildew Plant Village SVM- 83.26% ResNet152V2. 98.01%
Classification

12 Fusarium Wilt Plant Village SVM 92.75% M-CNN 97.13%


13 Anthracnose Plant Village SVM 92.75% CNN 97.13%
14 Downy Mildew plant village Random Forest 97.65% CNN 96.08%
and Decision
Tree
15 Stem Diseases Plant Village SVM 74.3% VGG-16 97.89%

Fig-1. accuracy Results from representation in Graph.


III. Methodology

According to the Fig-2. The leaf is identified and applied to the segmentation for
disease variation of clusters and Testing is performed. If testing fails then a trial on images
is going to be applied and further processed out to the next level for both testing and tail
images by histogram. Reproduction is given to CNN then it is digitally tracked by
converting it into frequency from time domine using DFT (Discrete Fourier
transformation). Both the inputs are given to google net then it is forwarded to tracking
input after that it is forwarded pooling to Softmaxing. This gives predicting output with the
input dataset.

Fig-2. Proposed Chili Leaf disease detection methodology

IV. Experimentation

Fig-3-Public Dataset of diseased Chili leaves


IV.I. Image Pre-processing

Figure 3 shows how images in the dataset are downsized to 128 by 128 resolutions to speed up
training and make model testing calculations more realistic. By minimizing the input or goal
variables, the optimization technique aids in training processing speed. While protecting the
image's data from loss of integrity

Fig.4: Pre-Processed Chilli Leaves


IV.II. Training Dataset

In chilies, primarily three types of disease classes are trained for disease detection, with the
training data size specified by Mosaic (80) and Gemini (80), and Healthy (80) images of each
class. A screenshot of the training Dataset with two diseased classes and one Healthy class is
shown in the figure.

Fig-5- A screenshot of the training Dataset with Three Diseases.

IV.III. Training Classes

Figure- 5. depicts two types of training results: Training Validation Accuracy and Training
Data Loss functions. The training dataset employs multi-layer Deep convolution networks with
five pooling layers, ten convolution layers, and ten Relu-convolution layers, as well as three types
of popular training optimizer algorithms, including the Stochastic Gradient Descent with
Momentum (SGDM) optimizer, the Adam optimizer, and the Root Mean Squared Propagation
optimizer (RMSProp). By varying the number of epochs used by the aforementioned three
algorithms, the best accuracy for the Chili disease dataset can be found. Other training parameters,
such as Sequence length is longest, Initial Learn Rate is 0.01, Minibatch Size is 128, Sequence
Padding Value is 0, Sequence Padding Direction is "right," L2-Regularization is 0.0001,
LearnRateDropFactor is 0.1, LearnRate Additionally, the L2norms, which measure the separation
between a vector coordinate and the vector space's origin, are computed using the Gradient
Thresholding Method.
V. Results and Discussion
Based on the training results in Table 2anf Fig-8 and Fig-9 the training data achieved 98.61%
accuracy with the Google Net architecture by taking into account the previously specified
parameters, where training accuracy varies depending on the optimizer algorithm and epochs
used. Simulations revealed that the training data set achieved 98.61% accuracy in two cases, one
with the ADAM optimizer producing 98.61% accuracy with 40 epochs and the other with the
SGDM. But the SGDM optimizer produces the same 98.61% accuracy in three epochs which are
20,30,40 epochs.
The RMSPROP optimizer did not perform well for the Google Net on the Chili disease
dataset, as the maximum training accuracy is only 75.00%. However, when Max Epochs are set to
20, SGDM achieves 98.61% accuracy, while ADAM achieves it in 40 Max Epochs. Screenshots
of training accuracy test results using SGDM and ADAM optimizer are shown in Figures 7 and
8. When SGDM and ADAM are compared, the SGDM performance is superior because 98.61%
training accuracy was achieved in only 20 Epochs. As shown in figures 9 and 10, lower and
higher epochs resulted in lower accuracy in SGDM and ADAM, while only middle-range epochs
achieved higher accuracy.

Fig.6: Screenshot of Training accuracy test results using SGDM optimizer-Learning rate-0.01 and Epoch-20
and Accuracy-98.61%
Fig.7: Screenshot of Training accuracy test results using ADAM optimizer-Learning rate-0.01 and
Epoch-40 and Accuracy-98.61%

Table.2: Training Data Performance based on Optimizer and Epochs

Optimizer Algo. Max_Epochs Accuracy (%)


10 65.28%
20 98.61%
SGDM 30 98.61%
40 98.61%
50 33.33%
10 40.28%
20 66.67%
ADAM 30 83.33%
40 98.61%
50 68.06%
10 33.33%
20 33.33%
RMSPROP 30 63.89%
40 75.00%
50 54.17%
Figure-8: Achievement of SGDM optimizer

Figure-9: Achievement of ADAM optimizer

VI. Conclusion
India is a heavily agricultural nation, where 60% of the people rely on farming for their
living. Reduce plant leaf diseases as soon as possible to increase agriculture productivity in the
nation. In the southern region of India, the yield of the chili crop, one of the most prolific crops,
has been declining over the past few years due to diseases. The goal of the current research is to
identify the infection in the chili plant at an early stage, which will assist farmers in taking quick
action. This study explains how Google Net layers can be used to identify diseases in chili leaf
tissue. Along with the usage of 69 Google Net layers, different training parameters are also
employed, first with constant values, and later with changing optimizer and epoch settings.

By acquiring data on the training accuracy for the various optimizers, including SGDM,
ADAM, and RSMPROP, an experiment is carried out. With Max Epochs of 50,40,30 and 20,
respectively, ADAM and RMSPROPF outperformed the other two optimizers in terms of
training correctness, both achieving 98.61%. Therefore, compared to ADAM and SGDM, the
SGDM optimizer produced the best accuracy with a Max Epochs value of 20. Last but not least,
it can be inferred from the experimental results that the Google Net layers perform best with
SGDM in order to attain 98.61% accuracy for smaller datasets.

REFERENCES

[1]. Sushil R. Kamlapurkar, “Detection of Plant Leaf Disease Using Image Processing Approach,”
International Journal of Scientific and Research Publications, Vol. 6, Issue 2, February 2016.
[2]. Sachin D. Khirade et al., “Plant Disease Detection Using Image Processing,” 2015, pp. 768-771, 2015.
DOI: 10.1109/ICCUBEA.2015.153.
[3]. Zulkifli Bin Husin et al., “Feasibility Study on Plant Chili Disease Detection Using Image Processing
Techniques,” 2012, pp. 291-296, 2012. DOI: 10.1109/ISMS.2012.33.
[4]. R. Meena Prakash et al.,” Detection of Leaf Diseases and Classification using Digital Image Processing”.
(ICIIECS).2017. DOI: 10.1109/ICIIECS.2017.8275915.
[5]. Pranjali B. Padol et al.,” SVM classifier based grape leaf disease detection”. IEEE Xplore:17 November
2016, INSPEC Accession Number: 16482445. DOI: 10.1109/CASP.2016.7746160
[6]. [6]. Baldomero Manuel Sanchez Rangel et], al “KNN-based image segmentation for grapevine potassium
deficiency diagnosis”. Accession Number: 15872438. in 2016.
DOI: 10.1109/CONIELECOMP.2016.7438551.
[7]. Aakanksha Rastogi et al,” Leaf disease detection and grading using computer vision technology & fuzzy
logic”. Accession Number: 15077184,2015. DOI: 10.1109/SPIN.2015.7095350.
[8]. Shivani K Tichkule, et, al “Plant diseases detection using image processing techniques”. Accession
Number: 16864613.2016. DOI: 10.1109/GET.2016.7916653.
[9]. C.S. Sumathi et al,” Neural Network based Plant Identification using Leaf Characteristics Fusion”. Volume
89 – No.5, March 2014. DOI:10.5120/15499-4141.
[10]. Boran Sekeroglu et al “Leaves Recognition System Using a Neural Network”.
doi.org/10.1016/j.procs.2016.09.445.
[11]. Asha Patil & Kalpesh Lad et,al. 05 January 2022. Feature Selection for Chili Leaf Disease Identification
Using GLCM Algorithm, pp 359–365.Springer. DOI: 10.1007/978-981-16-3945-6_35.
[12]. Muhammad Alyas Khan1, Mushtaq Ali,20 August 2021, Machine Learning-based Detection and
Classification of Walnut Fungi Diseases. IASCvol.30, no.3, pp. 771–785, DOI:10.32604/iasc.2021.018039.
[13]. A. Anagnostis, et, al. 8 January 2020. A Convolutional Neural Networks Based Method for Anthracnose
Infected Walnut Tree Leaves Identification, Athanasios Anagnostis. Volume10 ,Issue
2 .10.3390/app10020469 . DOI.org/10.3390/app10020469
[14]. Md. Aashiqui Islam 1, et.al. 28 December 2020, Machine Learning Learning-basedassification of Papaya
Disease Recognition. ISBN: 978-1-7281-6386-4. DOI: 10.1109/ICECA49313.2020.9297570.
[15]. Benjamin Doh, et.al. 2019. Automatic citrus fruit disease detection by phenotyping using machine learning.
In IEEE, DOI: 10.23919/IConAC.2019.8895102.
[16]. Kshyanaprava Panda Panigrahi, et, al. 27 march 2020. Maize Leaf Disease Detection, and Classification
Using Machine Learning Algorithms. (AISC, volume 1119). DOI: 10.1007/978-981-15-2414-1_66.
[17]. Tin yr Wiesner-Hanks1, et.al. 12 December 2019. Millimeter-level Plant Disease Detection from Aerial
Photographs via Deep Learning and Crowdsourced.Data. pp-1-
11.https://doi.org/10.3389/fpls.2019.01550.
[18]. Ashutosh Kumar Singh, 1 SVN Sreenivasu 2 U.S.B. K. et.al. 2022. Hybrid Feature-Based Disease
Detection in Plant Leaf Using Convolutional Neural Network, Bayesian Optimized SVM, and Random
Forest. | Article ID 2845320.pp-1-16 | https://doi.org/10.1155/2022/2845320.
[19]. Karcher Ahmed, et.al. 16 April 2020. Rice Leaf Disease Detection Using Machine Learning Techniques
Syed Md. Irfanul Alam and Sifat Momen. DOI: 10.1109/STI47673.2019.9068096.
[20]. S. Ramesh, D. Vydeki, et.al. June 2020, Recognition and classification of paddy leaf diseases using
Optimized Deep Neural network with Jaya algorithm. Volume 7, Issue 2, Pages 249-260. Pp-1-
12.https://doi.org/10.1016/j.inpa.2019.09.002.
[21]. Bharti Pate, Aakanksha Sharaff, et.al. 2022. Review on phenotype traits analysis of Rice plants by using
Machine Learning Techniques. 2022.pp-1-32.DOI: https://doi.org/10.21203/rs.3.rs-1444587/v1.
[22]. Varun Pramod Bhartiya et.al. 26 May 2022. Rice Leaf Disease Prediction Using Machine Learning. pp-
DOI: 10.1109/ICPC2T53885.2022.9776692.
[23]. Feng Jiang, Yang Lu, et.al. December 2020, Image recognition of four rice leaf diseases based on deep
learning and support vector machine. Volume 179, 105824
https://doi.org/10.1016/j.compag.2020.105824.
[24]. Himani Bhatheja; N. Jayanthi, et.al.16 February 2022.Detection of Cotton Plant Disease for Fast
Monitoring Using Enhanced Deep Learning Technique.DOI: 10.1109/ICEECCOT52851.2021.9708045.
[25]. U. Shruthi; V. Nagaveni, et.al. 06 June 2019. A Review on Machine Learning Classification Techniques for
Plant Disease Detection. DOI: 10.1109/ICACCS.2019.8728415.
[26]. Asha Patil and Kalpesh Lad, et.al. 02 October 2020. Chili Plant Leaf Disease Detection Using SVM and
KNN Classification. (AISC, volume 1187). DOI: 10.1007/978-981-15-6014-9_26.
[27]. R. Toscano-Miranda, et. al. 23 May 2022.Artificial intelligence and sensing techniques for the management
of insect pests and diseases in cotton: a systematic literature review. pp. 16 –
31.DOI: https://doi.org/10.1017/S002185962200017X.
[28]. Anil Bhujel 1,2, Na-Eun Kim, et.al. 4 February 2022. A Lightweight Attention-Based Convolutional Neural
Network for Tomato Leaf Disease Classification. Pp-1-18. https://doi.org/10.3390/agriculture12020228.
[29]. Uday Pratap Singh, et, al. 27 March 2019. Multilayer Convolution Neural Network for the Classification of
Mango Leaves Infected by Anthracnose Disease. DOI: 10.1109/ACCESS.2019.2907383.
[30]. Ke Lin1, Liang Gong1, et.al. 15 February 2019. Deep Learning-Based Segmentation and Quantification of
Cucumber Powdery Mildew Using Convolutional Neural Network.pp-1-10.
https://doi.org/10.3389/fpls.2019.00155.
[31]. Suffolk et.al. 28 May 2021.Symptom-Based Identification of G-4 Chili Leaf Diseases Based on Rotation
Invariant. Volume-8. Article-650134 - https://doi.org/10.3389/frobt.2021.650134.
[32]. Syed Barges Zehra1, Asif Ahmad2 et.al. Published in Sept.-Oct., 2017. Chilli Leaf Curl Virus an Emerging
Threat to Chilli in India. IJPAB. ISSN: 2320 – 7051.pp-1-8. DOI: http://dx.doi.org/10.18782/2320-
7051.5471.
[33]. Al-bay, et.al. 2020. Evolutionary Feature Optimization for Plant Leaf Disease Detection by Deep
Neural Networks. Vol. 13(1),2020. pp.12–23.DOI: https://doi.org/10.2991/ijcis.d.200108.001.
[34]. Arsenic, M., Karanovic, et.al. 2019. Solving current limitations of deep learning-based approaches for plant
disease detection. Volume 11, Issue 7,10.3390/sym11070939.https://doi.org/10.3390/sym11070939.
[35]. Anand, R., Veni, S. et.al. 2016. An application of image processing techniques for the detection of diseases
on brinjal leaves using the k-means clustering method. DOI: 10.1109/ICRTIT.2016.7569531.
[36]. Abdullah, N. E., Hashim, et.al. 2012. A characterization of watermelon leaf diseases using Fuzzy Logic.
September). DOI: 10.1109/ISBEIA.2012.6422869.
[37]. Alias, N., Nashat, S. 2011, November.Classification gel electrophoretic image of DNA Fusarium
graminearum featuring support vector machine. DOI: 10.1109/ICSIPA.2011.6144122.
[38]. Al Bashish, D. Break. et.al, 2010, December. A framework for detection and classification of plant leaf and
stem diseases. DOI: 10.1109/ICSIP.2010.5697452.
[39]. Balakrishna, K., & Rao, M. et.al. 2019. Tomato plant leaves disease classification using KNN and
PNN. (IJCVIP), DOI: 10.4018/IJCVIP.2019010104.
[40]. Bajpai, G., Gupta, A. et.al.2019, July. Real-Time Implementation of Convolutional Neural Network to
Detect Plant Diseases Using Internet of Things. (CCIS, volume 1066). Pp- 510–522.DOI: 10.1007/978-
981-32-9767-8_42.
[41]. Barbados, J. G. A. et.al, 2017. Digital image processing techniques for detecting, quantifying, and
classifying plant diseases. (2013). SpringerPlus, 2(1), 660. Vol. 9, No. 1. https://doi.org/10.1186/2193-
1801-2-660.
[42]. Antoni Buades; Triet M. Le.et.al.2010. Fast cartoon+ texture image filters. IEEE Transactions on Image
Processing,19(8), 1978-1986. DOI: 10.1109/TIP.2010.2046605.
[43]. Costa, J., Silva, C. et.al. 2019.Hierarchical Deep Learning Approach for Plant Disease Detection. pp 383–
393 (LNIP, volume 11868). DOI: 10.1007/978-3-030-31321-0_33.
[44]. Jayraj Chopda; Hiral Raveshiya.et,al.2018.Cotton Crop Disease Detection using Decision Tree Classifier.
DOI: 10.1109/ICSCET.2018.8537336.
[45]. Andre da Silva Abade1, et.al.2019 Plant Diseases Recognition from Digital Images using Multichannel
Convolutional Neural Networks. ISBN: 978-989-758-354-4.DOI: 10.5220/0007383904500458.
[46]. de Luna, R. G., et.al.2018 Automated image capturing system for deep learning-based tomato plant leaf
disease detection and recognition. DOI: 10.1109/TENCON.2018.8650088.
[47]. Doh, B., Zhang, et.al.2019. Automatic Citrus Fruit Disease Detection by Phenotyping Using Machine
Learning. In 2019 25th International Conference on Automation and Computing (ICAC) (pp. 1-5). IEEE.
DOI:10.23919/IConAC.2019.8895102.
[48]. Durmuş H, et.al.2017. Disease detection on the leaves of the tomato plants by using deep learning. In 2017
(pp. 1-5). IEEE. DOI: 10.1109/Agro-Geoinformatics.2017.8047016.
[49]. Konstantinos P,et.al. 2018.Deep learning models for plant disease detection and diagnosis. Computers and
Electronics in Agriculture, Volume-145, Pages-311-318. https://doi.org/10.1016/j.compag.2018.01.009.
[50]. Fuentes, A., Yoon, S., et.al.2017. A robust deep-learning-based detector for real-time tomato plant diseases
and pests’ recognition. Volume 17 -Issue 9- Sensors.https://doi.org/10.3390/s17092022.
[51]. Francis, J., & Anoop, B. K.et, al.2016. Identification of leaf diseases in pepper plants using soft computing
techniques. Volume 5 -Issue 3. https://doi.org/10.3390/bios5030537.
[52]. Fang, Y., & Ramasamy, R. P. et.al. 2015.Current and prospective methods for plant disease detection.
Biosensors, 5(3), 537-561.pp-1-25. https://doi.org/10.3390/bios5030537.
[53]. Fadzil, W. N. W. M., Rizam, et.al.2014. Orchid leaf disease detection using border segmentation
techniques. DOI: 10.1109/SPC.2014.7086251.
[54]. Geetharamani, G., & Pandian, A.et, al. June 2019. Identification of plant leaf diseases using a nine-layer
deep convolutional neural network.Volume-76, Pages-323-
338https://doi.org/10.1016/j.compeleceng.2019.04.011.
[55]. Gavhale, K. R., Gawande, et.al. 2014, April. Unhealthy region of citrus leaf detection using image
processing techniques. DOI: 10.1109/I2CT.2014.7092035.
[56]. Gulhane, V. A., &Kolekar, M. H.et, al. In 2014.Diagnosis of diseases on cotton leaves using principal
component analysis classifier. DOI: 10.1109/INDICON.2014.7030442.
[57]. Gibson, G. J., Otten, et.al. 2006. Bayesian estimation for percolation models of disease spread in plant
populations. Stat Comput 16, 391–402 (2006). DOI 10.1007/s11222-006-0019-z.
[58]. Hossain, E., Hossain, et.al. In 2019. Color and texture-based approach for the detection and classification
of plant leaf disease using KNN classifier. DOI: 10.1109/ECACE.2019.8679247.
[59]. Iqbal, Z., Khan, M.et. al. 2018.Automated detection and classification of citrus plant diseases using image
processing techniques. https://doi.org/10.1016/j.compag.2018.07.032.
[60]. Jaisakthi, S.M., Mirunalini.et.al. In 2019 February. Grape Leaf Disease Identification using Machine
Learning Techniques. DOI: 10.1109/ICCIDS.2019.8862084.
[61]. Julia, M., Kumar, A.et. al. In 2013. Image processing for smart farming: Detection of disease and fruit
grading. DOI: 10.1109/ICIIP.2013.6707647.
[62]. Kumari, C. U., Prasad, et.al. In 2019.Disease Detection: Feature Extraction with K-means clustering and
Classification with ANN. DOI: 10.1109/ICCMC.2019.8819750.
[63]. Kusumo, B. S., Haryana, et.al.2018. Machine learning-based for automatic detection of corn-plant diseases
using image processing. DOI: 10.1109/IC3INA.2018.8629507.
[64]. Khiladi, S. D. et.al. 2015.Plant disease detection using image processing.
DOI: 10.1109/ICCUBEA.2015.153.
[65]. Khan, J. F., Adhami, R. Ret. al.2008.Color image segmentation utilizing a customized Gabor filter.
DOI: 10.1109/SECON.2008.4494353.

You might also like