Tomato disease detection using deep learning convolutional neural network

Tomato leaf disease detection and
classification
based on deep convolutional neural
networks
Priyanka Pradhan and Dr. Brajesh Kumar
Department of Computer Science & IT, MJP Rohilkhand
University, Bareilly-243006, India

Content
• Outlines of Key Principles
• Literature Survey
• Results and findings
• Conclusion
• References
4/6/2022 2

Outlines
1. Tomato is one of the major crops in India. The production of tomatoes is
rigorously affected by several types of diseases. Therefore, initial
detection of disease is vital for the quality and quantity of tomatoes.
2. There are many diseases that mostly affect the plant leaves. This paper
adopts a convolutional neural network (CNN) model to detect and
identify diseases using the images of tomato leaves.
3. The proposed CNN model comprises four convolutions and four max
pooling layers, which are followed by fully connected layers. The
performance of the proposed method is assessed by performing
experiments on a well-known PlantVillage dataset.
4. The overall accuracy of the proposed method is obtained as 96.26%.
5. It is compared with some fine-tuned pre-trained CNN models
InceptionResNetV2 and InceptionV3.
6. The results illustrate that the proposed method outperforms all the
methods based on fine-tuned models.
4/6/2022 3

Tomato Dataset
Figure 1: Different Tomato Healthy and Unhealthy Class
4/6/2022 4
The performance of the proposed method is assessed by performing
experiments on a well-known PlantVillage dataset. There are nine diseases and
one healthy class for tomato crop in the dataset.

Dataset for Image Classification
4/6/2022 5
PlantVillage consists of 54303 images for 14 different crops distributed in 38
categories including healthy and various disease classes. The total number of
tomato images considered for this work was 14,532 which includes both
infected tomato leaves, as well as healthy tomato leaves images The 75%
images are kept in training set and the rest 25% images from the testing set.
Table I

Literature Survey
[1] Karthik et al. introduced two dissimilar deep convolutional neural networks (CNN) models
to identify the infection in tomato leaves in as early stage as possible. The authors stated overall
accuracy of 98%.
[2] Agrawal et al. Created a model entailing three convolutional, three max pooling layers,
and two fully connected layers. Test results showed 91.2% accuracy.
[3] Ashok et al. developed a method that identifies tomato leaf disease based on image
segmentation and clustering to classify leaf disease of tomato plants.
[4] Ahmad et al. tested four pre-trained CNN models VGG-16, ResNet, VGG-19, and
InceptionV3 to identify and classify tomato leaf diseases. They tested these four models on two
datasets and found that InceptionV3 gave the best performance.
[5] Elhassouny et al. developed an effective smart mobile application based on CNN for
detecting and classifying tomato leaf diseases.
[6] Hasan et al. used the transfer learning concept to reorient the Inception model to classify
tomato leaves in three disease classes.
[7] Sardogan et al. developed a CNN both for feature extraction and classification using spectral
information.
[8] Tm et al. achieved the results similar to various well known techniques using optimal
computing resources.
4/6/2022 6

Work flow of Proposed System
Figure 2: Work flow for tomato leaf disease detection
4/6/2022 7
The proposed workflow for CNN-based plant disease detection
and identification is shown in Figure 2. This flowchart shows
data preparation module, dataset spliting module and
classification module.

Classification Framework
Figure 3: Classification framework for plant disease detection and identification
4/6/2022 8
CNN classification framework used in proposed work consists of
four convolutional layers and four max pooling layers placed
alternatively. At the top of the arrangement of convolutional and
pooling layers a fully connected layer is placed. The well-known
RELU activation function is used at different layers in CNN. At the
top layer in the model, Softmax classifier is used to map the input to
appropriate class as shown in Figure 3.

Different layers used in Proposed CNN
Model
4/6/2022 9
Table II
The implemented CNN model layer details are given in Table II. For
minimizing the training loss, the Adaptive Moment Estimation optimizer is
used. The learning rate is set to 0.001 and batch size is kept as 32 during the
training process.

Results and Findings
4/6/2022 10
The naive accuracy values obtained for the proposed model and
two pre-trained models InceptionResnetV2 and InceptionV3 are
listed in Table III along with learning error and training time. Both
class-wise and overall accuracies are given in the table. The
training time is reported in seconds.

Contd…
4/6/2022 11
Table IV provides the classification accuracy in terms of Cohen’s Kappa.
Cohen’s Kappa statistic is a very good measure that can handle very well
both multi-class and imbalanced class problems.
Cohen’s Kappa= (Overall Accuracy – Expected Accuracy)/ (1-Expected Accuracy)

Training Loss vs Training Accuracy
Graph of Proposed Model
4/6/2022 12
Figure 4: Training Loss vs Training Accuracy Graph of Proposed Model

CONCLUSION
1. In this work, a CNN model was developed for the automatic
detection and identification of tomato crop diseases using leaf images.
2. It consists of four convolution, four max pooling layers, and three
fully connected layers with varying numbers of filters at different
layers.
3. The experimental results exhibited that the average accuracy of the
model is 96.21%.
4. It was able to detect different tomato diseases with good accuracy.
5. The efficiency of the proposed model was found better and takes a
lesser training time than pre-trained CNN models Inception- ResnetV2
and InceptionV3. Also, the proposed model showed good accuracy as
compared with other models.
4/6/2022 13

REFERENCES
[1] R. Karthik, M. Hariharan, S. Anand, P. Mathikshara, A. Johnson, and R. Menaka,“Attention embedded
residual CNN for disease detection in tomato leaves, ”Applied Soft Computing, vol. 86, pp. 105933, 2020.
[2] M. Agarwal, A. Singh, S. Arjaria, A. Sinha, and S. Gupta,“ ToLeD: Tomato leaf disease detection using
convolution neural network, ”Procedia Computer Science, vol. 167, pp. 293-301, 2020.
[3] S. Ashok, G. Kishore, V. Rajesh,S. Suchitra, S. G. Sophia, and B. Pavithra, “Tomato Leaf Disease Detection
Using Deep Learning Techniques, ” In 2020 5th International Conference on Communication and Electronics
Systems (ICCES), pp. 979-983, IEEE, 2020.
[4] I. Ahmad, M. Hamid, S. Yousaf, S. T. Shah, and M. O. Ahmad, “ Optimizing Pretrained Convolutional
Neural Networks for Tomato Leaf Disease Detection, ” Complexity, 2020.
[5] A. Elhassouny, and F. Smarandache, “ Smart mobile application to recognize tomato leaf diseases using
Convolutional Neural Networks, ”In 2019 International Conference of Computer Science and Renewable
Energies (ICCSRE), pp. 1-4, IEEE, 2019.
[6] M. Hasan, B. Tanawala, and K. J. Patel,“ Deep learning precision farming: Tomato leaf disease detection by
transfer learning, ”In Proceedings of 2nd International Conference on Advanced Computing and Software
Engineering (ICACSE).
[7] M. Sardogan, A. Tuncer, and Y. Ozen, “ Plant leaf disease detection and classification based on CNN with
LVQ algorithm, ”In 2018 3rd International Conference on Computer Science and Engineering (UBMK), pp.
382-385, IEEE, 2018.
4/6/2022 14

Tomato disease detection using deep learning convolutional neural network

More Related Content

What's hot (20)

Similar to Tomato disease detection using deep learning convolutional neural network (20)

More from Priyanka Pradhan (19)

Recently uploaded (20)

Tomato disease detection using deep learning convolutional neural network

Editor's Notes