Plant Leaf Disease Detection

Rohit R. Kanade1, a *, Sonam N. Bakkamr2,b , Hrishikesh B. Kadam3,c And Bhavesh S. Sanap4,d

1
At Post Kolad,Raigad, India
2
At Post Mangaon,Raigad, India
3
At Post Mahad,Raigad,India
4
At Post Kolad,Raigad,India
a
[email protected], [email protected], [email protected],
b
d
[email protected]

Keywords: convolutional neural network,cnn,deep learning,leaf diseases identification,

Abstract. In this paper, we proposed a plant leaf disease detection system using a convolution neural
network (CNN). The CNN model is trained using dataset with 5 different classes of plant. The
proposed model achieves 94% Classification accuracy. This accuracy is greater than the accuracy of
machine learning approaches. The proposed model is also tested with respect to its consistency and
reliability.

Introduction

In this proposed system we used a CNN model to solve the Plant Leaf Detection problems. The CNN
is a class of deep learning algorithm. There are different convolutions that are performed in several
layers of CNN model. There are four types of layers in CNN model: Conv_2d layer, Max_pooling
Layer, Flatten Layer and dense layer.
Here we use image processing to detect the diseases. We can put the image into a system and a
computer can perform various phases for identification and detect the related classes to which that
image belongs. This work aims to make a leaf recognition technique based on the specific
characteristics derived from images. We classify the images into four categories depending on the leaf
diseases. For training, phase loads the variety of images and it resizes with a resolution. Then images
and corresponding labels are appended to the list. The above steps are used for testing the data.
For classification here, we use the CNN algorithm. It consists of four layers for efficient
implementation. In each step, convolutional layer build and pooling are added. Finally, the flatten layer
is added to get the inputs and create 1D output. After that the dense layer added to produce the output
as dot product of input and kernel. After the model building, load the data in the model. Here we use
the variable consisting of a model name that represents healthy or diseased. Then save the model on
the folder to this variable name and put the data to this model and detect it.

Proposed System

Our proposed system detects the plant diseases of different classes with the accuracy of 95%. Our
project builds with minimal design so everyone can use it for their work. In our proposed system it
also detects the healthy leaf by uploading image.

Existing System

In most of the areas Farmers use traditional methods, like sorting out the disease from their past
experience, or getting information from the internet. These methods may not always lead to perfect
result. Also, there are agricultural institutes which are available to the farmers at all time. Scientists
will visit the fields and take the samples of the diseased plants for testing. The results will take some
time, but during this time some damage might already might be done. And when it comes to a new
type of disease, the damage might be even worse. There are also case that all the villages may have
the facility of Local administration offices, to report the information to the nearest institute. Thus, the
problem cannot be identified Apart from traditional methods there are also advanced methods that
are already implemented.

Machine Learning approaches include traditional computer vision algorithms like haar, hog, sift,
surf, image segmentation, Support Vector Machines (SVM), using K-Nearest Neighbours (KNN), K-
means and Artificial Neural Networks(ANN).Deep Learning based plant disease classification
models includes the use of a variety of CNN models such as AlexNet, GoogleNet, VGGNet etc. It
seems that sometime the availability of datasets for some crops may not be available and sometime
the available dataset may not be sufficient.

Literature Survey

1. Smart Farming: Pomegrante Disease Detection Using Image Processing, 2015

A web-based tool has been developed to identify fruit diseases by uploading fruit image to the A web-
based tool has been developed to identify fruit diseases by uploading fruit image to the system.
Feature extraction has been done using parameters such as colour, morphology and system. Feature
extraction has been done using parameters such as colour, morphology and CCV (colour coherence
vector). Clustering has been done using the k-means algorithm. SVM CCV (colour coherence vector).
Clustering has been done using the k-means algorithm. SVM is used for classification as infected or
non-infected. This work achieved an accuracy of 82% is used for classification as infected or non-
infected. This work achieved an accuracy of 82% to identify pomegranate disease. [1]

2. Leaf Disease Detection and Recommendation of Pesticides using Convolution Neural

Network,2018

Crop production problems are common in India which severely effect rural farmer, agriculture sector
and the country’s economy as a whole. In Crops leaf plays an important role as it gives information
about the quantity and quality of agriculture yield in advance depending upon the condition of the
leaf. In this paper we proposed the system which works on pre-processing, feature extraction of leaf
images form plant village dataset followed by convolution neural network for classification of disease
and recommending pesticides using Tensor flow technology. The main two processes that we use in
our system is android application with java Web Services and Deep Learning. We have use
Convolution Neural Network with different layers five, four & three to train our model and android
application as a user interface with JWS for interaction between these systems. Our result show that
the highest accuracy achieved for 5-layer model with 95.05% for 15 epochs and highest validation
accuracy achieved for 5-layer model with 89.67% for 20 epochs using tensor flow. [2]

3. An Artificial Intelligence and Cloud Based Collaborative Platform For Plant Disease
Identification, Tracking and Forecasting for Farmers, 2018

This paper presents an automated, low cost and easy to use end-to-end solution to one of the biggest
challenges in the agricultural domain for farmers-precise, instant and early diagnosis’ of crop diseases
and knowledge of disease outbreaks-which would be helpful in quick decision making for measures
to adopted for disease control. This proposal innovates on known prior art with the application of
deep Convolutional Neural Networks (CNN) for disease classification, introduction of social
collaborative platform for progressively improved accuracy, usage of geocoded images for disease
density maps ad expert interface for analytics. High performing deep CNN model “Inception” enables
real time classification of diseases in the Cloud platform via a user facing mobile app. [3]

4. Plant Leaf Diseases Detection and Classification Using Image Processing and Deep Learning
Techniques, 2020

This paper presents a system that is used to classify and detect plant leaf diseases using deep learning
techniques. The used images were obtained from plat village dataset website. In our work, we have
taken specific types of plants; include tomatoes, pepper, and potatoes, as they are the most common
types of plants in the world and in Iraq in particular. This Data Set Contains 20636 images of plants
and their diseases. In our Proposed system, we used the convolutional neural network (CNN),
Through which plant leaf diseases are classified 5 classes. As a result, we obtained excellent accuracy
in training and testing, we have got an accuracy of 95% for all data set that were used. [4]
Methodology
The complete processes of designing, training and validating the model for plant disease identification
using the CNN are described further in detail. The entire method is separated into a number of stages
in the following subsections, starting with collecting the images for the classification process.
1. Data acquisition and pre-processing
The diseased and healthy plant leaf images were captured by team. The dataset containing 1500
images of 3 different plant leaves Mango, Guava and Rice. In pre-processing we visualize the images
and then we reseized them. we have resized the image size of each classes into 256x256 to get better
accuracy and avoid overfitting issue. Then we convert those images into a Numpy array and normalize
them because the CNN model does not take image as input. After that we visualize the class count
for class imbalance. and we have taken same image quantity of each classes to avoid dataset
imbalance issue. then we splitting the dataset into the train, validate and test sets.

Fig. 1 Image Plotting

2. Training the CNN Model
Once pre-processing is done, then we started training the CNN Model. The CNN Model build with
the help of TensorFlow and keras..By using this model, we classify the image that the system is
getting after pre-processing of testing image Then we get particular disease name or status of healthy
leaf if there is no disease on that leaf. The model includes various layer that are required for classify
the results.
2.1 Conv_2d layer -
It is used to create convolutional kernel that is convolved with the input layer to produce the
output tensor.
2.2 Max_pooling Layer -
It is a downsampling technique which takes out the maximum value over the window defined
by poolsize.
2.3 Flatten Layer -
It flattens the input and crates a 1D output.
2.4 Dense Layer -
Dense layer produces the output as the dot product of input and kernel.

Fig.2 Proposed CNN Model

We can see in above figure we have created network architecture for the model. We have used
different types of layers to their features namely conv_2d,max_poolling,flatten and dense layer. First,
we use conv_2d layer with the filter size of 32. As we can see the set the input shape as (256,256,3)
because we reshaped the train test data in particular size. Then we added layer called maxpooling
layer and we set the pool size 3. Then again, we added a conv_2d layer with filter size of 16. After
that flatten layer added for creating the 1D output. Then we added two dense layers first contain 8
output neurons and second contain 5 because we have taken 5 disease classes an we use softmax
activation because we are doing the categorical classification. At the end we got total 231365 trainable
parameters.

Fig.3 Model Training

In above figure fitting the model with data and finding out the accuracy at each epoch to see how our
model is learning. Now we will train our model on 75 epochs and a batch size of 128. We can try
using a greater number of epochs too increase accuracy but here we can see that the model has already
reached a very high accuracy so we don’t need to run it for more. During each epoch we can see how
the model is performing by viewing the training and validation accuracy.

Fig.4 Model Accuracy

In above figure we can see the accuracy of the model. Red lines are indicating the training accuracy
and blue lines are indicates the validation accuracy. As we can see accuracy started increasing around
at 15 epochs and keep increasing till getting the higher accuracy. In our case we got accuracy above
90%,
Conclusion
Deep learning is a efficient technique for image processing and pattern recognition and it can
effectively solve the problems in the identification of plant leaf diseases. The proposed CNN model
can effectively classify 5 different classes of healthy and diseased plants using leaf images. The
proposed model achieves and average accuracy of 94% in the classification of the testing set plant
leaf images. The number of training epochs, batch size had greater impact on the respective results.
The max pooling performs much better than average pooling. Compared with other machine learning
models, the proposed CNN modes have better ability and performance.
Acknowledgement
We would sincerely like to thank our guide for this project Prof.K.R.Metha for providing us her
valuable time and support throughout the project. We would also like to extent our gratitude to
Prof.K.R.Metha (Head of Department of Computer Engineering) and all the other faculty members
for helping us generously. We would like to thank Teaching & Non-teaching staff of Computer
Department who helped us time to time in all respects. We acknowledge with thanks, the assistance
provided by department staff, library, and lab attendants.
References

[1] M. Bhange and H. Hingoliwala, "Smart Farming: Pomegranate Disease Detection Using
Image," Second International Symposium on Computer Vision and the Internet, p. 9, 2015.

[2] P. K. Kosamkar, . V. Kulkarni, K. Mantri, S. Rudrawar, S. Salmpuria and N. Gadekar, "Leaf

disease detection and recommendation of pesticides using convolution neural network," fourth
international conference on computing communication control and automation , p. 4, 2018.

[3] K. K. Singh, "An Artificial Intelligence and Cloud Based Collaborative Platform for Plant,"
IEEE International Conference, p. 8, 2018.

[4] M. A. Jasim and J. M. AL-Tuwaijari, "Plant Leaf Diseases Detection and Classification,"
International Conference, p. 7, 2020.