Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 26

PDEC0201 Research Seminar

Classify Mango Leaf Diseases using modified CNN


Ravi Anand
Part time-PhD
Roll No-205EC025

Under the supervision of


Dr. Ritesh Kumar Mishra
Dated-15-12-2023
Department of Electronics and Communication Engineering
National Institute of Technology, Patna
Outlines
1. Introduction
2. Research Objective & Goal
3. Motivation of the Research
4. Data set and methodology
4.1 Data Set
4.2 Proposed Methodology
4.3 Flow chart
4.4 Algorithm
5. Approach Use
5.1 Performance Evaluation
6. Results
7. Conclusion
8. Publications
9. References
1. Introduction
• The mango tree is severely afflicted by a fungus called anthracnose,
most noticeable on the fruits and leaves. This paper's primary goal is
to provide a viable system for an early and affordable solution to this
problem.
• This paper proposed a tool based on modified convolutional Neural
Network was developed to identify mango leaf diseases. Modified
CNN has shown excellent performance on recognition and image
rating issues.
2. Research Objective & Goals

To develop a model and/or a tool for Detection and diagnosis of


mango leaf diseases using modified CNN General mango leaf
diseases detection system and comparative study on machine
learning classification techniques for mango leaf disease
detection.

Use of customize CNN with the help of KNN Classifier to


improve the classification results for finding high accuracy and
detects more number of diseases of multiple crops.
3. Motivation of the Research
• Farmer selects the suitable crop based on type of soil, weather condition
of the location and economic value.
• The agriculture industries started searching new methods to increase
production of food because of increasing population, changes in weather
and instability in politics. This makes researchers to search new efficient
and precise technologies for high productivity.
• The diseases in crop are creating problem of low production and
economic losses to farmers and agricultural industries. Therefore
identification of disease and its severity based as become necessary
4. Data set and Methodology

• The data set we take for this research is from Kaggle


https://www.kaggle.com/datasets/dalipkamboj/mango-leaf-specie
s-indian
).
• Dataset contains Mango Leaves image samples which are
divided into three categories named as healthy, Anthracnose and
Black Soothy Mold. For this study three class samples are
considered and further these sample images are augmented, and
similar type of images are generated through different
augmentation techniques
4.2. Proposed Methodology
• The proposed methodology comprised of three main section:
(i) pre-processing section
(ii) Feature extraction using CNN and (iii) classification using KNN classifier .

Figure 1: Proposed model for mango leaf disease detection


Work flow of Proposed Method

Figure2: Working of proposed model for mango leaf disease detection


Flow chart
• a)The detection process initialized with input images(𝑰𝑹𝑮𝑩 ). from image 216
database infected images.
• b) Some preprocessing is done on resize image size(256*256)
• c) Feature Section is done by basic algorithm of CNN.
• d) Classification is done with modification of CNN with KNN Classifier.
• e) The infected part is segmented (𝑵𝒎𝒂𝒔𝒌) from input image (𝑰𝑹𝑮𝑩 )
• f) The mask images (𝑵𝒎𝒂𝒔𝒌) are further sub-divided into sub-images 𝑵𝒔𝒖𝒃.
• g) These 𝑵𝑹𝑶𝑰 are used a Region of Interest (RoI), which identifies leaf disease.
• h) Performance evaluation is done in terms of Accuracy, Sensitivity, 226 Specificity
4.3. Algorithm

• Input: Input Infected Color Images of mango leaf (𝑰𝑹𝑮𝑩 )


• Output: Leaf Disease Detection.
• a) Evaluate masking 𝑵𝒎𝒂𝒔𝒌 from (𝑰𝑹𝑮𝑩 ).
• b) 𝑀𝑚𝑎𝑠𝑘 Into smaller regions𝑵𝒔𝒖𝒃. ;
• c) for (𝑵𝒔𝒖𝒃 in 𝑁𝑚𝑎𝑠𝑘) do
• d) Classify 𝑵𝒔𝒖𝒃 into 𝑵𝑹𝑶𝑰 Mango leaf diseases.
• e) if 𝑵𝑹𝑶𝑰 has infected, then categorise disease
• f) end
5. Performance Evaluation

• Performance evaluation of the model is done by different Metrics


such that the model's accuracy (Acc) is evaluated as:
𝐴𝑐𝑐 = (𝑇𝑃+𝑇𝑁)/ (𝑇𝑃+𝑇𝑁+𝐹𝑃+𝐹𝑁 )
• Sensitivity (Sen) is evaluated as: 𝑆𝑒𝑛 = 𝑇𝑃/ 𝑇𝑃+𝐹N
• Specificity (Spec) is evaluated as: 𝑆𝑝𝑒𝑐 = 𝑇𝑁 /𝑇𝑁+𝐹𝑃
6.Results: Comparative analysis
MODEL ACCURACY SENSTIVITY SPECIFICTY
kNN 0.6607 0.6541 0.6748

SVM 0.8141 0.8145 0.8176


Neural Network 0.7774 0.7775 0.7797
(Proposed Model) 0.921436 0.999504 0.859114

• It has been observed from the above Table that KNN classifier yields 97.30 % of accuracy, 97.30 % of
precision and 97.30 % of recall. Similarly, CNN yields 98.92 % of accuracy, 98.91 % of precision and
98.90 % of recall. Although, proposed model yields 99.37 % of accuracy, 99.36 % precision and 99.35
% recall.
• Therefore, it can be concluded that the proposed model (hybrid model) achieved outstanding accuracy
for the classification of mango leaf disease identification.
Results: Confusion matrix for the proposed model

From the obtained confusion matrix it can be noted that the out of 608, 605
test instances of healthy class are correctly detected and remaining 3
instances are misclassified as anthracnose and soothy mold.
Among 1508 samples of anthracnose class, 1498 samples are correctly
identified and remaining 10 misclassified samples are identified as healthy
(7) and soothy mold (3).
Similarly, 1207 samples of BS Mold class are correctly identified and
remaining 10 samples are misclassified as anthracnose class and healthy
class.

Fig 3: Obtained confusion matrix for the proposed model


Results: ROC Curve at K folds 10 and 20

Fig4: ROC Cure at K Fold 10 Fig5: ROC Curve at K Fold 20

The performance of the proposed model is also plotted in terms of ROC curve. The obtained ROC Cure at
K Fold 10 and ROC Cure at K Fold 20 is shown in Fig4 and Fig5 respectively.
Result
7. Conclusion
• .

The system's most powerful technique is to identify illnesses of various


plants in an agricultural setting, where the disease's key features
include speed and accuracy.
Tool based on proposed 270 algorithm gives accuracy level 92.1436 %,
sensitivity 99.9504% and specificity 85.9114%.
Levels of accuracy, sensitivity and specificity which is higher than any
other state of the art algorithm
8.Publication

Link of Submitted Paper


9.References
• [1] Sharma, M. T. S. S., & Kumar, H. (2022). Detection and
Classification of Plant Diseases by Alexnet and Googlenet Deep
Learning Architecture.
• [2] Arivazhagan, S., Shebiah, R. N., Ananthi, S., & Varthini, S. V.
(2013). Detection of unhealthy region of plant leaves and
classification of plant leaf diseases using texture features. Agricultural
Engineering International: CIGR Journal, 15(1), 211-217.
• [3] Iqbal, Z., Khan, M. A., Sharif, M., Shah, J. H., ur Rehman, M. H., &
Javed, K. (2018). An automated detection and classification of citrus
plant diseases using image processing techniques: A review.
Computers and electronics in agriculture, 153, 12-32.
References
• [4] Patil, J. K., & Kumar, R. (2011). Advances in image processing for detection
of plant diseases. Journal of Advanced Bioinformatics Applications and
Research, 2(2), 135-141.
• [5] Camargo, A., & Smith, J. S. (2009). An image-processing based algorithm
to automatically identify plant disease visual symptoms. Biosystems
engineering, 102(1), 9-21.
• [6] Sankaran, S., Mishra, A., Ehsani, R., & Davis, C. (2010). A review of
advanced techniques for detecting plant diseases. Computers and electronics
in agriculture, 72(1), 1-13.
• [7] Trongtorkid, C., & Pramokchon, P. (2018, February). Expert system for
diagnosis mango diseases using leaf symptoms analysis. In 2018 International
Conference on Digital Arts, Media and Technology (ICDAMT) (pp. 59-64). IEEE.
References
• [8] Singh, U. P., Chouhan, S. S., Jain, S., & Jain, S. (2019). Multilayer
convolution neural network for the classification of mango leaves infected
by anthracnose disease. IEEE Access, 7, 43721-43729.

• [9] Arya, S., & Singh, R. (2019, September). A Comparative Study of CNN and
AlexNet for Detection of Disease in Potato and Mango leaf. In 2019
International Conference on Issues and Challenges in Intelligent Computing
Techniques (ICICT) (Vol. 1, pp. 1-6). IEEE.

• [10]Tumang, G. S. (2019, July). Pests and Diseases Identification in Mango


using MATLAB. In 2019 5th International Conference on Engineering,
Applied Sciences and Technology (ICEAST) (pp. 1-4). IEEE
References
• [11] Ullagaddi, S. B., & Raju, S. V. (2017, January). Disease recognition
in Mango crop using modified rotational kernel transform features. In
2017 4th International conference on advanced computing and
communication systems (ICACCS) (pp. 1-8). IEEE.

• [12] Y.-N. Wang, L. Jin, and H. Mao, “Farmer cooperatives’ intention


to adopt agricultural information technology-mediating effects of
attitude,” Information Systems Frontiers, vol. 21, no. 3, pp. 565–580,
2019.
References
• [13]Sahni, V., Srivastava, S., & Khan, R. (2021). Modelling techniques to
improve the quality of food using artificial intelligence. Journal of Food
Quality, 2021.

• [14]Khan, R., Dhingra, N., & Bhati, N. (2022). Role of Artificial Intelligence
in Agriculture: A Comparative Study. In Transforming Management with
AI, Big-Data, and IoT (pp. 73- 317 83). Springer, Cham.

• [15]Khan, R., Kumar, S., Dhingra, N., & Bhati, N. (2021). The Use of
Different Image Recognition Techniques in Food Safety: A Study. Journal
of Food Quality, 2021.
References
• [16] Katiyar, S., Khan, R., & Kumar, S. (2021). Artificial Bee Colony Algorithm for Fresh
Food Distribution without Quality Loss by Delivery Route Optimization. Journal of Food
Quality, 2021.
• [17] Khan, R., Tyagi, N., & Chauhan, N. (2021). Safety of food and food warehouse using
VIBHISHAN. Journal of Food Quality, 2021.
• [18] J. L. Fernandez, D. P. Losada, and R. Sanz, “Enhancing building security systems
with autonomous robots,” in Proceedings of the 2008 IEEE International Conference on
Technologies for Practical Robot Applications, Woburn, MA, USA, November 2008.
• [19 ]T.-D. Cao and H.-L. Truong, “Analyzing and conceptualizing monitoring and
analytics as a service for grain warehouses,” Recent Developments in Intelligent
Information and Database Systems, vol. 642, pp. 161–171, 2016. 331
• [20] S. I. Lao, K. L. Choy, G. T. S. Ho, R. C. M. Yam, Y. C. Tsim, and T. C. Poon, “Achieving
quality assurance functionality in the food industry using a hybrid case-based reasoning
Review Of Paper
•Some other plant leaves can be studied.
•Abstract needs to be revised.
•How your method is better than existing method.
•Results are not properly discussed.
•Results and conclusion should be revised.
I thank to the Honorable DSC members and Guides for their humble
co-operation.
I would like to make use of this platform for suggestion for
improvement the strength of this study.

Thank You Sir


RAVI ANAND
Thank
You . ……

You might also like