Detection of Plant Disease FF

Detection of plant disease using SVM and Neural
Network
Shyam gupta , Shubham mittal , Arjun singh
ABSTRACT :
Disease in plants plays a very important role in overall growth and
development of plants . There are many factors which can cause disease in
plants like external environment , lack of nutrition etc . Whenever a plant
gets any disease it directly impacts its leaves and fruits. Different diseases
cause different types of patterns on leaves. Usually remote sensing
technology is used to find multi spectral image and hyper spectral images .
By taking images and processing through different pattern recognition
algorithms we can detect the disease. The major concern is accuracy , to
achieve accuracy new method is proposed which combine BP, PCA are
combined with SVD methods
INTRODUCTION :
Image processing is a new technology . There are lots of uses of this
technique in different sectors . images are of two types analog and digital .
For any image processing we need an image in digital format, for
digitization the image should be examined on each sample are quantised
utilised a range of bits . Then the processing of images is done by
computers .[1] With back propagation , we can train a network covering a
whole range of layers .[3] Good options networks are marked as feed
forward networks. Their related patterns coordinate and form a non cyclic
graph, when the whole network is prepared for a specific implementation
,then that network is transformed to be trained . now training or learning
can start.[2]
Learning techniques are mainly of two types: Supervised learning and
unsupervised learning . Supervised learning is defined when there is a fixed
labelled data set and the model is getting trained according to that data
set[4] . In unsupervised learning technique there is no labelled data set for
training the model ,system learn by itself, here information neither
classified nor labelled .[5]
LITERATURE REVIEW :
In various research different methods have been introduced for plant

detection disease. With this paper , we are giving a different technique for
detection of disease in plant leafs classifying .The k -Nearest-Neighbors
(kNN) algorithm is used to predict the class of a single test . Possibly this
may be the most easy algorithm for a single test exemplar . A prominent
hindrance of the kNN strategy is that it requires a large amount of training
data. Neural networks are used to hinder data and require less
computational resources. but it is tough to understand the algorithm
structure in a neural network. Support Vector mechanism (SVM) was used
to get the optimised reachable Machine learning algorithms arranged in
high dimensional sets of data. In SVM quantitative complexity is very low
due to quadrilateral improvement issues which make it easy to control
complexity of decision principle and number of error frequencies. There are
some drawbacks of support vector mechanism (SVM) which are deciding of
best suitable parameters that suits the data sets becomes very hard when
training information is not linearly divisible .[1]
The topic of this paper is on how to identify plant disease at an early stage
and accurately , by using the various kinds of image processing technology
along with artificial neural networks (ANN) .The foundation that is
designed here is for plant disease acknowledgment , by kipping the
framework in mind continuous optimization being done . Here in this
research paper we have applied various techniques of image processing
along with artificial neural networks to detect disease in plants in the early
stage . We have achieved an accuracy rate of 91% which is good
improvement in results from previous for that we noticed that Gabor
channel for extraction highlight and for classification showed sign of
improvement we use artificial neural network (ANN) . [3]
For automatic disease detection image processing can be used . The study
was conducted to identify the disease spots on the leaves of "Monocot" and
"Dicot" plants using colour segmentation. Various classification techniques
were used in this study, including CIELAB , HSI and YCBCr colour space .
The results showed that CIELAB , HSI and YCbCr colour space are useful in
identifying disease spots on leaves of monocot and dicot plants . [5]
Plant disease, or disease of plants, is a complex phenomenon that has been
studied for decades. Reasons for disease in plants are bacteria, fungi and
different kinds of viruses. Plant diseases are spread by insects, birds,
mammals (including humans) and other animals. The most common types
of diseases are fungal infections and bacterial infections. Here in this
research paper , different types of plant disease will be discussed with the
help of various ways of image processing technology so that it becomes easy
to detect disease in plants at an early stage and automatically without any
medical process. Here we will also talk about the impacts on food chain
quality of plant productions etc . [7]
Support vector mechanism (SVM)
Support vector mechanisms (SVM) is a powerful machine learning tool

that is now used by researchers around the world to solve a variety of
problems. They have been used to classify plants based on their leaf shape
and size, as well as their ability to photosynthesize at different wavelengths
of light.In this paper, we will describe our research into support vector
machines and how they can be applied to plant disease classification. We
will also discuss some of the challenges that researchers face when using
SVM models for such tasks and what kinds of improvements might be
possible if these issues could be resolved
Previous studies have shown that SVM can be used to classify data.
However, these studies have focused on linearly separable data, where two
classes are represented as linearly separated feature vectors. When dealing
with non-linearly separable data (e.g., plant leaves), it is important to
remember that there may be multiple relationships between the variables
that define each class. For example, there may be more than two classes
present in the dataset.
The methods used to research this topic included:
- Various classification techniques such as Decision Trees and Naive Bayes
Classifier
- Kernel function such as radial basis function (RBF), Sigmoid Foundation
Polynomial Kernel Function, and Gaussian Kernel Function
- Parameter Tuning on each method was performed by evaluating their
ability to correctly classify new samples while minimising misclassification
error rate
PROPOSED METHODOLOGY :
Here we are using three types of algorithm,these are:

A. Back-propagation Algorithm (BPA) :
It is a method for training neural networks that uses the error signal to
adjust the weights in the network. It's a popular technique because it's
computationally cheap, and it can be used on many different kinds of
networks, including multilayer and fully connected.
The idea behind back-propagation is simple: after each iteration of

calculating the error signal, we take the gradient of that error signal and use
it to modify our weights. This process is repeated over and over until we
reach a steweight update .
B. Principal Component Analysis (PCA) :
Principal component analysis is a type of unsupervised learning algorithm
which is used for reduction of dimensionality in machine learning . Along
with orthogonal transformation some numerical processes are used to
convert the observation of correlation features into a set of linear uncorrelated
features . After this new transformation this feature is called the principal
component .
C. Singular Value Decomposition (SVD) :
Singular Value Decomposition (SVD) is a reliable and vigorous / strong
orthogonal matrix decomposition method .It is used to factorization the
matrix in three different matrix due to singular value decomposition image
processing became more and more optimised and popular in the single
processing area . Single value decomposition has many important
properties in image processing but its main properties are :
- SVD is good for transformation of algebra for processing of image
-SVD has important properties in imaging
D. HSI COLOUR SPACE
HSI COLOUR SPACE model is an upgrade of the RGB model . The HSI
model is very much the same as a human vision. It senses the colour the
same as humans . The HSI was introduced to better manipulation of
colour and was developed to closely replicate the human perspective of
colour sensing . The HSI is very suitable for images processing tasks like
equalization ,convolution , histogram etc . It works by changing the
brightness value since I depend on R,G and B. H always gives values
ranging between 0 to 180 degrees . [10]
But H can have values between 0 to 360 degrees. If G<B , then the degree
of H will be more than 180 degrees . so , if G<B , we only have to compare H
first then take it as the hue value .
SOME COLORS AND THREE HSI INTENSITY VALUE
COLOUR H S I
GREEN 60% 1 0.375
BLUE 120% 1 0.375
RED 180% 1 0.375
BLACK 60% 1 0.375
WHITE - 0 0.75
YELLOW - 0 0
Flow chart :
Figure 1
The first thing you need to do is load your training data image and your test
image.
Then, we use an algorithm called Extreme Detection to find the positions of
the objects in each image. We compare those positions with the positions of
the objects in our training data photos . Then for sparse feature
representation systems generate descriptors which have both character
detection and extraction .
The resistance point in the feature is to find the same photo for another
photo , so we have to continuously keep matching between two photos .
Now both images will be put together one at a 90 degree angle and another
at a different angle but the position of both images will be the same . After
all this if the image is still not clearly seen so change the angle then a
filtering process is introduced which filters images /object/ pixels in image
then again process of matching starts where keys points and image are
being compared .
EXPERIMENTAL RESULTS :
In figure 2, numbers of photos are taken of data and thus photos are
labelled as training data sets. The second folder named the train data set
has some other photos . The train and trained data are then matched using
the SIFT algorithm. [9]
Figure 2:Healthy Part of leaf
Figure 3: disease parts of leaf

In figure 3 , by using support vector mechanism disease can be detect and
that leaf can be distinguished , then by enhance SVM we will find how
much percent of leaf which are infected .[12]
A b
Figure 3 : histrogram equalization Image (a) before contrast enhancement , (b) after
contrast enhancement
CONCLUSION :
In this study we conducted to detect the plant diseases in grape and wheat
using BP networks. The two types of diseases were detected by using three
colour features, colour, texture and shape. Image recognition was
conducted by using PCA to diminish the dimensions of the features data
set . By using the introduced methodology disease can be found in plants
in the early stage and can be cured in the early stage so that disease will not
harm the whole tree.
In the below table ,the ratio of the number of correctly identified samples
to the total number of tested samples is called accuracy . Here values in
accuracy percentage of each disease are shown .
REFERENCES :
[1]. Savita N. Ghaiwat, Parul Arora, “Detection and Classification of Plant Leaf Diseases Using
Image processing Techniques: A Review”, International Journal ofRecent Advances in
Engineering and Technology, ISSN(online) :2347-2812, Volume 2 Issues 3 2014
[2]. Prof. Sanjay B. Dhaygude and Mr.Nitin P.Kumbhar, “Agricultural plant Leaf Disease
Detection Using Image Processing” International Journal of Advanced Research in
Electrical,Electronics and Instrumentation Engineering Vol. 2, Issue , January 2013
[3]. Anand.H.Kulkarni and Ashwin Patil R. K, “Applying image processing technique to detect
plant diseases”, International Journal of Modern Engineering Research (IJMER), Vol.2,
Issue.5, Sep-Oct. 2012 pp-3661-3664 ISSN: 2249-6645 2012
[4]. Haiguang Wang, Guanlin Li, Zhanhong Ma, Xiaolong Li, “Image Recognition of Plant
Diseases Based on Backpropagation Networks”, 5th International Congress on Image and
Signal Processing (CISP 2012) 2012
[5]. Piyush Chaudhary Anand K. Chaudhari Dr. A. N. Cheeranand Sharda Godara “Color
Transform Based Approach for Disease Spot Detection on Plant Leaf”, International Journal of
Computer Science and Telecommunications [Volume 3, Issue 6, June 2012
[6]. Rani Pagariya and Mahip Bartere, “ Review paper on identification of plant diseases using
image processing technique”, 2014
[7]. Jayamala K. Patil and Raj Kumar, “ Advance in image Processing for detection of plant
disease”, Journal of Advanced Bioinformatics Applications and Research ISSN 0976-2604 Vol
2, Issue 2, June-2011, pp135-141
[8]. Prof. Sanjay B. Dhaygude and Mr.Nitin P.Kumbhar, “ Agricultural Plant Leave detection
using image Processing”, International Journal of Advanced Research in Electrical, Electronics
and Instrumentation Engineering Vol. 2, Issue 1, January 2013
[9]. Yan K, Rahul Sukthankar , “pca-sift: a more distinctive representation for local image
descriptors” Computer Vision and Pattern Recognition, 2004, CVPR 2004.,Proceedings of the
2004, IEEE Computer Society Conference on ,2004 pp.506-513.
[10]. Kim, D. G., T. F. Burks, J. Qin, and D. M. Bulanon, “Classification of grapefruit
peeldiseases using color texture feature analysis”, International Journal on Agriculture and
Biological Engineering, 2(3): 41-50, 2009
[11]. Arivazhagan, S., R. N. Shebiah, S. S. Nidhyanandhan, and L. Ganesan, “Classification
ofcitrus and non-citrus fruits using texture features”, Computing Communication and
Networking Technologies,ICCCNT-2010.
[12]. Quraishi, M.I.; Choudhury, J.P.; De, M., "Image recognition and processing using
Artificial Neural Network," Recent Advances in Information Technology (RAIT), 2012 1st
International Conference on , vol., no., pp.95,100, 15-17 March 2012

Detection of Plant Disease FF

Uploaded by

Copyright:

Available Formats

Detection of Plant Disease FF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Detection of Plant Disease FF

Uploaded by

Copyright:

Available Formats

Detection of plant disease using SVM and Neural

In various research different methods have been introduced for plant

Support vector mechanisms (SVM) is a powerful machine learning tool

Here we are using three types of algorithm,these are:

The idea behind back-propagation is simple: after each iteration of

SOME COLORS AND THREE HSI INTENSITY VALUE

Figure 2:Healthy Part of leaf

Figure 3: disease parts of leaf

You might also like