COTTON LEAF RECOGNITION AND DISEASE

PREDICTION WEBAPP
Yogeshwar khandagre
Yogkhandagre444@gmail.com
Mrs. Sherin Eliyas
sherine@hindustanuniv.ac.in
Department of Computer Applications
Hindustan Institute of Technology and Science, Padur chennai
Abstract—Agriculture is very important part for any
country and we can say that India is considered to be an
agro-based country. India ranks second in the agricultural
output worldwide. Agriculture helps to boosts various
industries such as food industry and textile industry, and
in textile industry cotton producing crops have a special
position amongst all other textile crops. And we can say
that Cotton plant is important cash crops in India. And
have there are a variety of diseases on cotton plant that
have had a negative impact on the performance of cotton
plants over the past few years. The disease is diagnosed
by farmers too late and is very difficult and expensive to
control. And because of disease loss productivity of
cotton. For more profit and good productivity its intense
care is necessary. The aim of the system is to develop an
web based application which can recognizes and predict
cotton leaf diseases. Our Goal Is build a website where
we can upload an image of any cotton plant leaf and this
website it will tell you information about decease .cotton
plant have any decease or not or its burn by fertilizer.
Cotton farmer need to upload the image and then with
the help of image processing we can get a digitized color
image of a diseased leaf and then we can proceed with
applying CNN to predict cotton leaf disease. Many
researchers are using machine learning for early
detections of cotton plant disease. Convolution neural
network (CNN) is a deep feed forward artificial neural
network. This algorithm is little faster as compared to
other classification algorithms. In this paper, CNN is
used for classification of the diseased portion of cotton
plant images. 
Keywords— agriculture production with AI, AI based
website, Deep Convolution neural network , Artificial
neural network
1.Introduction
Agriculture has played an important role in the financial
development of India and most another agricultural
countries. India ranks second in the agricultural output
worldwide. In India most of the land is used for agriculture
purpose. It is also the back support for Indian farmer’s
financial condition. The most important requirement is the
acceptance of the infection and its planning. huge amount
of loss in quality and quantity of cotton yield due to
different diseases affecting the plant which obstruct the
growth of crops in fields which may cause huge loss in the
quality of products. The existing method for plants disease
detection is simply naked eye observation which requires
more man labor, properly equipped laboratories, expensive
devices etc and some method is expensive. Nowadays image
processing is used a lot for detecting such diseases, Pests
like the germ, fungus, and microorganisms.
This paper is based on Artificial Intelligence system
especially Deep learning. As Farmer, we know Farmer can’t
solve Farm’s complex and even small problems due to lack
of perfect education. So as AI enthusiastic we decided to
solve this problem using the latest technology like AI. In
which implements Convolutional Neural Network to detect
cotton leaf diseases. Cnn is easy to use for discover
infection created by Bacteria and environmental effects. In
early stage Disease detection is challenging task for farmers
where physical presence is a must. Disease detection and
recognition is the very important. There are various
algorithms in image processing for disease recognition by
image classification like KNN, SVM, Random Forest,
Artificial Neural Network and CNN. We all know many
Farmer can’t solve Farm’s complex and even small
problems due to lack of perfect education. The goal of this
application is to develop a system which recognizes crop
diseases. Our Goal Is build a website where we can upload
an image of any cotton plant leaf and this website it will tell
you information about which decease it is on plant and how
we can treat plant decease.
A. Purpose of Proposed System:
1. Developing a user-friendly web-based system for
farmers.
2. Recognizing Cotton leaf diseases accurately from
input images.
3. Providing corrective and preventive measures for
the detected diseases
4. Provide disease information
B. Cotton leaf diseases focused:
1. Bacterial Blight.
2.Literature Review
In recent year, limited number of works has been done
related to cotton disease classification, cotton stage
Identification using deep learning models.
According to “Mahesh Shivaji Dange et.al.”[1] Infected leaf
characteristics are compared to normal leaf texture features.
Improved processing has four main steps, first the RGB
color image creation is created, and then the RGB image is
converted to HIS because RGB is for color reproduction and
HIS for color definition. Then the green pixels are hidden
and removed using a certain amount of limit, then the image www.plantvillage.org).The content has been written by
is split and the useful parts are extracted, finally the texture plant pathology experts, reflecting information is sourced
statements are calculated using spatial Gray Level from the scientific literature.
Dependence Matrices.
Another approach based on leaf images and using ANNs as
According to “Tushar J.Haware”( 2015) [2] an algorithm a technique for an automatic detection and classification of
that works well with high precision mixing is used to detect plant diseases was used in conjunction with -means as a
plant diseases. K-means integration is used for partitioning clustering procedure proposed by the Z. H. Zhou and S. F.
and the method of computational computing. The red- Chen, “Neural network ensemble,” [8]. ANN consisted of
colored, red-blue Zeros pixels and pixels on the boundary of 10 hidden layers. The number of outputs was 6 which was
the infected group are removed. the number of classes representing five diseases along with
the case of a healthy leaf. On average, the accuracy of
According to [3], author Mrunalini represents the technique classification using this approach was 94.67%.
to classify and identify the different diseasethrough which
plants are affected. In Indian Economy a Machine learning According to S.Vijay et al [9] analyzed the cotton leaf
based recognition system will prove to be very useful as it disease and applied the image enhancement and k means
saves efforts, money and time too. The approach given in clustering techniques. Image processing technique is easy
this for feature set extraction is the color co-occurrence and accurately for leaf disease.
method. For automatic detection of diseases in leaves,
neural networks are used. The approach proposed can
significantly support an accurate detection of leaf, and 3.METHODOLOGY
seems to be important approach, in case of steam, and root
In this procedure of developing the model for cotton plant
diseases, putting fewer efforts in computation.
disease recognition using deep CNN .system classifies the
leaf image using image classification algorithm CNN.
According to S.Batmavady, S.Samundeeswari [4] focused
A Convolutional Neural Network (ConvNet/CNN) is a Deep
on applying image processing and neural network
Learning algorithm which can take in an input image, assign
techniques to identify the diseased cotton leaf. The leaf
importance (learnable weights and biases) to various
images are taken from village plant dataset. Segmentation is
aspects/objects in the image and be able to differentiate one
done using FCM techniques and classification is done by
from the other. The pre-processing required in a ConvNet is
RBF Neural network.
much lower as compared to other classification algorithms.
While in primitive methods filters are hand-engineered, with
According to Namrata R.Bhimte, V.R.Thool [5] established
enough training, ConvNets have the ability to learn these
image processing approaches for automatic detection of
filters/characteristics. The whole process of training a model
cotton leaf spot disease. Classification, feature extractions
for plant disease detection using CNN in-depth is described
like color, texture of image using SVM classifier and
in detail. The complete process is divided into several
various preprocessing steps were done and colour based
required sections in the steps below:
segmentation is used to obtain the segmented part of leaf
spot disease.
A. Project Description:
According to Ranjith et al.[6] provide smart irrigation
The model for plant disease recognition using deep
system has been proposed which can control the irrigation
learning CNN . The complete process is divided into some
automatically using an android mobile application. Apart
stages, Appropriate datasets are required at all stages of
from this, the photos of plant leaves are captured and are
object recognition research, starting from training phase to
sent to the cloud server, which is further processed and
evaluating the performance of recognition algorithms.
compared with the diseased plant leaf images in the cloud
Images of the infected plants are captured by digital camera
database. Based on the comparison a list of plant diseases
and processed using image growing ,image segmentation
suspected is given to the user via the Android mobile
techniques to detect infected parts of the plants.
application.

According to [7] they incorporated all the hybrid features of

a. Image Acquisition:
a leaf color, texture shape (geometric feature) by the
respective methodology. Plant Village: a tool for crop In image processing, it is defined as the action of
health; an online platform dedicated to crop health and crop retrieving an image from some source, In this
diseases, called Plant Village (available at phase, raw image is taken as input from the user
 and converted into equivalent gray scale image.
Also the image is resized into size of 128*128.
b. convolutional Layers:
Convolutional layers apply a convolution operation
to the input, passing the result to the next layer.
A convolution converts all the pixels in its
receptive field into a single value. .The final output
of the convolutional layer is a vector. After the
alteration of captured image, the processed image
further passes through three different hidden layer
in which feature extraction, pooling and flattening
layer are also performed.
c. Disease Prediction:
After applying CNN, using Softmax layer the leaf
image is predicted with disease which is gaining
highest probability of occurrence.
B. Implementation:
Real time images are collected using web camera and stored
in database. Stored database images are converted into test
images and train images. In the Google drive, the upload
folder option is used to upload the test and train dataset.
Implementation steps used in Google colab (Google
colaboratory) is described below.
1. Dataset collection.
2. Uploading the dataset into drive.
3. Accessing colab.
4. Mounting drive in Google colab.
5. Coding implementation , testing and training.
1.1 Data set collection.
Real time images are collected using web camera and
stored in database. Stored database images are converted
into test images and train images.
1.2 Uploading the Dataset into Drive-
In the Google drive, the upload folder option is used to
upload the test and train dataset which are named as test
cotton and train cotton.
1.3 Accessing Colab-
* First, sign in the Google accounts.
* proceed to the Google colab welcome page.
* Click on the newpython3 notebook option to start the
session fresh.
* Select runtime menu option or notebook option to select
GPU.
* Configure notebook instance, to download the necessary
packages.
1.4 Mount Drive in Google Colab-
Click on mount drive option, an authorization code is
generated and it is entered in Google drive, which generates
the image folder path.
1.5 Code Implementation, Test and Train-
Python coding implementation based Python coding
implementation based on different CNN model like VGG16
and RESNET50. We should train the above CNN model and
test the images to get better accuracy and get the predicted
disease.
1. First, sign in the Google accounts. And proceed to
the Google colab welcome page.
2. Click on the newpython3 notebook option to start
the session fresh.
3. Select runtime menu option or notebook option to
select GPU.
Configure notebook instance, to download the necessary
pack Flask server is used to deploy the Web Application on
local computer. Server handles the request and response to
system use.
C. Algorithm:
CNN itself is a technique of classifying images as a part of
deep learning. In which we apply single neural network to
the full image.
i. Accepts a volume of sizeW1×H1×D1
ii. Requires four hyper parameters:
 Number of filters K
 Their spatial extent F
 The stride S
 The amount of zero padding P
iii. Produces a volume of size W2×H2×D2 where:
a. W2=(W1-F+2P)/S+1
b. H2=(H1−F+2P)/S+1(i.e. width and height
are computed equally by symmetry)
c. D2=K.
iv. With parameter sharing, it introduces F*F*D1
weights per filter, for a total of (F*F*D1)*K weights and K
biases. In the output volume, the dth depth slice (of size
W2*H2) is the result of performing a valid convolution of
the dth filter over the input volume with a stride of S, and
then offset by dth bias.
v. A common setting of the hyper parameters is F=3,
S=1,P=1
 First, sign in the Google accounts.
 And proceed to the Google colab welcome page.
 Click on the newpython3 notebook option to start
the session fresh.
 Select runtime menu option or notebook option to
select GPU.
 Configure notebook instance, to download the
necessary pacFlask server is used to deploy the Web
Application on local computer. Server handles the request
and response to system use
4.RESULTS AND DISCUSSION
The results training with the whole database containing both
original and augmented images. As it is known that Fig3.model loss
convolutional networks are able to learn features when
trained on larger datasets, results achieved when trained
with only original images will not be explored. 5. SYSTEM ARCHITECTURE
After fine-tuning the parameters of the network, an overall
accuracy of 96.3% was achieved, after the 100th training
iteration (95.8% without fine-tuning). Even after the 30th
training iteration high accuracy results were achieved with
exceedingly reduced loss, but after the 60th iteration, the
balance in accuracy and loss was carried out in high
accuracy.
the trained model was tested on each class individually.
Test was performed on every image from the validation set.
The results are displayed to emphasize how many images
from total of each class are accurately predicted.
Figure 8 illustrates trained model’s prediction results
separated for every class. The class numbers follow
enumeration from Table.

Fig2.Model accuracy

6. SOFTWARE INTERFACE DESIGN

 There are many methods in automated or computer vision
plant disease detection and classification process, but still,
this research field is lacking. In addition, there are still no
commercial solutions on the market, except those dealing
with plant species recognition based on the leaves images.
IN this system, a new approach of using deep learning
method was explored in order to automatically classify and
detect plant diseases from leaf images. New plant disease
image database was created, containing more than 1,100
original images taken from the available Internet sources
and extended to more than 30,000 using appropriate
transformations. The experimental results achieved
precision between 91% and 98%, for separate class tests.
The final overall accuracy of the trained model was 96.3%.
Fine-tuning has not shown significant changes in the overall
accuracy, but augmentation process had greater influence to
achieve respectable results.

9.References.

7. SOFTWARE WORKING SCREENSHOT 1. P. Rothe and R. Kshirsagar, “Adaptive

neuro-fuzzy inference system for recognition of
cotton leaf diseases,” in Computational Intelligence
on Power, Energy and Controls with their impact
on Humanity (CIPECH), 2014 Innovative
Applications of. IEEE, 2014, pp. 12–17.

2. A.-K. Mahlein, T. Rumpf, P. Welke et al.,

“Development of spectral indices for detecting and
identifying plant diseases,” Remote Sensing of
Environment, vol. 128, pp. 21–30, 2013.View
a. at: Publisher Site | Google Scholar.
a. Diseased cotton plant output

3. A.-K. Mahlein, T. Rumpf, P. Welke et al.,

“Development of spectral indices for detecting and
identifying plant diseases,” Remote Sensing of
Environment, vol. 128, pp. 21–30, 2013. View at
Publisher · View at Google Scholar · View at
Scopus.

4. S.Raj Kumar , S.Sowrirajan,” Automatic

b. Healthy cotton plant output
Leaf Disease Detection and Classification using
Hybrid Features and Supervised Classifier”,
International Journal of Advanced Research in
Electrical, Electronics and Instrumentation
Engineering, vol. 5, Issue 6,2016..

8. Conclusions
5. M. Francis and C. Deisy, "Disease Diseases from Paddy
Detection and Classification in Agricultural Plants
Using Convolutional Neural Networks — A Visual Plant Leaf Images”,
International Journal
Understanding," 2019 6th International Conference
on Signal Processing and Integrated Networks
(SPIN), Noida, India, 2019
of
6. ]Chaudhari Vaishnavi, Gondkar Sayali,
Shivarkar Pooja,”Survey on detection and
prediction of leaf diseases using CNN”, Indian
Journal of Automation and Artificial Intelligence,
Volume 6(10), 2019.

7. Patil Tushar, Palambe Shubham, Tawale

Gauri ,Patil Rajashree2sanchika Bajpai, “Cotton
Leaf Disease Identification Using Pattern
Recognition Techniques”, Multidisciplinary
Journal of Research in Engineering and
Technology, Volume 5, Issue 1,Pg.1-7,2018.

8. W. Xiuqing, W. Haiyan, and Y. Shifeng,

“Plant disease detection based on near-field
acoustic holography,” Transactions of the Chinese
Society for Agricultural Machinery, vol. 2, article
43, 2014.View at: Google Scholar.

9. R. Sarkar, A. Pramanik,” Segmentation of

Plant Disease Spots Using Automatic SRG
Algorithm: A Look Up Table Approach”, 2015
International Conference on Advances in Computer
Engineering and Applications (ICACEA) IMS
Engineering College, Ghaziabad, India.

10. K. Jagan Mohan, M. Balasubramanian,

“Detection and Recognition of Diseases from
Paddy Plant Leaf Images”, International Journal of
Computer Applications (0975 – 8887) Volume 144
– No.12, 2016.

11. K. Jagan Mohan,

M. Balasubramanian,
“Detection and
Recognition of

This template, modified in MS Word 2007 and saved as a
“Word 97-2003 Document” for the PC, provides authors
with most of the formatting specifications needed for
preparing electronic versions of their papers. All standard
paper components have been specified for three reasons: (1)
ease of use when formatting individual papers, (2) automatic
compliance to electronic requirements that facilitate the
concurrent or later production of electronic products, and (3)
conformity of style throughout a conference proceedings.
Margins, column widths, line spacing, and type styles are
built-in; examples of the type styles are provided throughout
this document and are identified in italic type, within
parentheses, following the example. Some components, such
as multi-leveled equations, graphics, and tables are not
prescribed, although the various table text styles are
provided. The formatter will need to create these
components, incorporating the applicable criteria that follow.
“Temperature (K)”, not “Temperature/K”.
ACKNOWLEDGMENT (Heading 5)
The preferred spelling of the word “acknowledgment” in
America is without an “e” after the “g”. Avoid the stilted
expression “one of us (R. B. G.) thanks ...”. Instead, try “R.
B. G. thanks...”. Put sponsor acknowledgments in the
unnumbered footnote on the first page.
REFERENCES
The template will number citations consecutively within
brackets [1]. The sentence punctuation follows the bracket
[2]. Refer simply to the reference number, as in [3]—do not
use “Ref. [3]” or “reference [3]” except at the beginning of a
sentence: “Reference [3] was the first ...”
Number footnotes separately in superscripts. Place the
actual footnote at the bottom of the column in which it was
cited. Do not put footnotes in the abstract or reference list.
Use letters for table footnotes.
Unless there are six authors or more give all authors’
names; do not use “et al.”. Papers that have not been
published, even if they have been submitted for publication,
should be cited as “unpublished” [4]. Papers that have been
accepted for publication should be cited as “in press” [5].
Capitalize only the first word in a paper title, except for
proper nouns and element symbols.
For papers published in translation journals, please give
the English citation first, followed by the original foreign-
language citation [6].
[1] G. Eason, B. Noble, and I. N. Sneddon, “On certain integrals of
Lipschitz-Hankel type involving products of Bessel functions,” Phil.
Trans. Roy. Soc. London, vol. A247, pp. 529–551, April 1955.
(references)
[2] J. Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd ed.,
vol. 2. Oxford: Clarendon, 1892, pp.68–73.
[3] I. S. Jacobs and C. P. Bean, “Fine particles, thin films and exchange
anisotropy,” in Magnetism, vol. III, G. T. Rado and H. Suhl, Eds.
New York: Academic, 1963, pp. 271–350.
[4] K. Elissa, “Title of paper if known,” unpublished.
[5] R. Nicole, “Title of paper with only first word capitalized,” J. Name
Stand. Abbrev., in press.
[6] Y. Yorozu, M. Hirano, K. Oka, and Y. Tagawa, “Electron
spectroscopy studies on magneto-optical media and plastic substrate
interface,” IEEE Transl. J. Magn. Japan, vol. 2, pp. 740–741, August
1987 [Digests 9th Annual Conf. Magnetics Japan, p. 301, 1982].
[7] M. Young, The Technical Writer’s Handbook. Mill Valley, CA:
University Science, 1989.
IEEE conference templates contain guidance text for
composing and formatting conference papers. Please
ensure that all template text is removed from your
conference paper prior to submission to the
conference. Failure to remove template text from
your paper may result in your paper not being published.
 /

We suggest that you use a text box to insert a graphic

(which is ideally a 300 dpi TIFF or EPS file, with all fonts
embedded) because, in an MSW document, this method is
somewhat more stable than directly inserting a picture.
To have non-visible rules on your frame, use the
MSWord “Format” pull-down menu, select Text Box >
Colors and Lines to choose No Fill and No Line.