Leaf Disease Detection and Pesticides Spraying System

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 4

LEAF DISEASE DETECTION AND PESTICIDES SPRAYING SYSTEM

NITHESH KUMAR K

M.Sc COMPUTER SCIENCE

SRI KRISHNA ARTS AND SCIENCE COLLEGE, COIMBATORE

Prof. S. DHANALAKSHMIM.Sc.,M.Phil

ASSISTANT PROFESSOR

SRI KRISHNA ARTS AND SCIENCE COLLEGE, COIMBATORE

COIMBATORE

ABSTRACT a smaller ecological footprint. But that's not all; the system
also includes a sophisticated pesticide spraying mechanism,
The diagnosis of plant diseases is crucial nowadays because
taking precision a step further. When a possible problem is
agriculture is the foundation of many countries, including
discovered, the system starts precise spraying, which only
India. Agriculture provides us with a variety of basic
treats the regions that need it. With this strategy, waste is
materials, particularly the crops that are the main source of
reduced and chemical abuse is avoided while still using
food for most people. However, there are two main causes of
pesticides efficiently. The result is a healthier surrounding
crop and plantation loss. The first is damage caused by
ecosystem, preserved biodiversity, and far lowers likelihood
natural disasters like floods, earthquakes, droughts, famines,
of pesticide resistance. There are many distinct kinds of plant
etc. The destruction caused by pathogens like viruses,
diseases, but the bulk of them fall into one of three groups:
bacteria, and fungi is the second factor. The type of disease
bacterial diseases, viral diseases, and fungi diseases. The
that plants are suffering from and how to protect them from it
classification followed by detection strategy is the best
are unknown to the farmer.
technique to find the disease. The basis for classification is
Keywords: Leaf Disease, K-Means Clustering, Support
the properties of shape and texture. Bacterial leaf spot is
Vector Machine
another name for this. Stone fruits, such cherry, plum, and
others, are most commonly affected by bacterial leaf spot. On
INTRODUCTION
various leaf parts, this disease causes black or dark patches to
The system's ability to detect diseases forms its core.
appear. This disease also exhibits yellow halos as a symptom.
It can accurately detect minute differences in leaf colour,
The size of the spots is erratic. Bacterial spots start appearing
texture, and structure - signs of potential health issues - by
on various parts of the top and bottom of the leaf, and if these
using high-resolution cameras and AI-driven picture
spots group together in one area, the illness will destroy that
processing. Farmers are better able to conduct targeted
area, killing the entire leaf. The development of the bacterial
treatment strategies thanks to these early warnings, which
illness in the plant leaves is also facilitated by moist and
eliminate the need for widespread pesticide application. In
chilly environments. Bacterial leaf spot can swiftly spread in
addition to preserving crop quality, this lowers the need for
these structures. Bacterial leaf spots typically appear on old
pesticides, resulting in better produce, financial savings, and
leaves, but they can also harm the tissues of young leaves.
Because viruses can spread across cells, viral infections leaf's margin. Scabs and wilting are the main signs of fungal
target the body from the inside out. Sometimes it's very disease. The end of the leaves turn yellow due to a fungus
difficult to diagnose viral diseases. Any part of a plant, assault on the plant's leaves.
including the leaves, roots, stem, and others, might be PROPOSED SYSTEM
affected by a virus. Green and yellow coloration can be
visible in leaves that have been infected by the virus, and
abnormal patterns can be recognised on the affected area. IMAGE ACQUISITION
For the purpose of creating the input dataset, image
The lifespan of a plant or any of its components that has
acquisition refers to gathering various types of samples. The
contracted a viral illness is extremely short. The productivity
various stages are then repeated for the dataset photos. A
and other elements are directly impacted. An image is made
dataset that includes the majority of the various types of
up of a number of pixels or dots that are kept in a rectangular
inputs is important in order to offer the best answer to any
array. Every every pixel has a unique type of colour. By
problem. We have a variety of plant leaves covered. In our
tallying up the number of pixels in an image, we may
dataset, various image formats have been captured. Our
determine its size. Both categories differ from one another.
approach may be applied to any other image outside of the
Each dot or pixel in a black-and-white image is either black
dataset as long as the size and format are suitable. The
or white, necessitating only one bit per pixel. Greyscale
classifier in our algorithm then predicts the disease for that
graphics, however, utilise 8 bits per pixel. Things become a
random image.
little more challenging for colour photographs. The height of
an image in colour images is defined as the number of bits at IMAGE PREPROCESSING AND ENHANCEMENT
each dot. It is additionally known as the bit plane. 2x colour
is feasible for bit planes made up of the symbol x. The colour It is the second stage of the processing of digital images.

information of an image can be stored using a variety of When utilising MATLAB as the input, image noise is

techniques. The RGB image, commonly known as a true minimised, pixel values are better categorised, spots are

colour image, is one of the methods. Red, green, and blue reduced, and contrast is improved. The goal of the picture

pixels are each represented by a three-dimensional array. pre-processing and enhancement stage is to make it simpler

The basic sign of these diseases is also a wrinkle on various to distinguish the diseased area from the surrounding area

parts of the leaves. Because each virus, if not adequately after the image has been pre-processed. Indirectly, pre-

treated, gives rise to a new variety of virus, its life span is processed images lacking contaminants perform better with

very large in comparison to the other types of disease, the classifier we apply. The pre-processed image additionally

making prompt disease prevention crucial. Another modifies the pixel values. The user can choose from a variety

prominent indication of these diseases is wrinkles on various of filters in MATLAB to improve the image.

parts of the leaves. Because each virus, if not adequately


treated, gives rise to a new variety of virus, its life span is
very large in comparison to the other types of disease,
making prompt disease prevention crucial. The fungi or other Figure 4.1: Image Preprocessing Original Image.

fungal organisms produce fungal illness. The ability of


fungus to propagate by wind and water is one of its
characteristics. Plants often exhibit grayish-green spots on
Figure 4.2: Pre-processed Image
their leaves, which if left untreated begin to spread to the
Figure 4.1 shows how the two photos differ from one used to create clusters. The closest candidate is chosen. We
another. The first image is the input image, and the second is employ the Euclidean shortest distance formula for this.
the pre-processed image where we used MATLAB to 7. The minimal distance from the centroid is used to group
improve contrast and other aspects of the image. Pre- the clusters
processing allows the user to reduce image noise and raises
the image's general quality. The second pre-processed image RESULTS AND DISCUSSION
separates the extra space from the original image as well.
Every pixel value increases the contrast for every individual
colour.
IMAGE SEGMENTATION
Figure Disease Detection Performance
We have three steps in our process, and the third is image Present the accuracy, precision, recall, F1-score, or other
segmentation. We divide the image into three clusters after relevant metrics of the disease detection algorithm used in the
image acquisition and pre-processing. Weasel stands for a system. Provide a comparison of the system's performance
color-based clustering algorithm used for clustering, and the against existing methods, if applicable. The system's ability
Otsu classifier is used for classification. Following the to detect diseases at an early stage, showcasing specific
creation of three distinct clusters, the user is prompted to examples where diseases were identified before visible
select one of the clusters, and the disease is then forecasted symptoms emerged. The effectiveness of the precision
using that specific cluster. spraying mechanism in delivering pesticides only to affected

K MEANS COLOR BASEDCLUSTERING areas. Quantify the reduction in pesticide use compared to
traditional blanket spraying. Highlight the system's capability
Euclidean shortest distance method. In this way different to monitor plants' health in real time and provide data on the
clusters are made. Implementation of the K mean algorithm frequency of monitoring and detection intervals. The
is as follows: alternaria alternate-infected plant leaf is loaded from the
1. Preprocessed image input database for analysis. In the second stage, contrast
2. Transform the specified colour space into the lab colour enhancement and image pre-processing are carried out. The
space. cluster's first loaded column in picture segmentation. The
3. The RGB colour space, which consists of the main colours disease, known as Alternaria, as well as the area of the
red, green, and blue, is the default colour space for the image. affected region in % are both displayed in the above figure.
4. In contrast to RGB colour space, it is simple to extract the The image is subjected to five hundred iterations in order to
various properties from lab colour space. test the accuracy of our intended methodology. Each time,
5. The L*a*b*, in which l stands for a sense of light be for the algorithm selects a different cluster, and accuracy is then
the blue/yellow value, and a for the red/green value, is predicted.
another way to express lab colour space. This colour space
offers a means of locating and communicating with the
colours. Around the a* and b*axes, all colour information is
located. CONCLUSION
6. K mean clustering is used to create several clusters based
on various colours. The distance between the initial The agriculture industry and other industries both

groupings and the distance of the other from that group are heavily rely on data mining tools. Data mining algorithms are
simple to implement, and we can use these methods to 7. A.Meunkaewjinda, P.Kumsawat and
resolve difficult issues. With the use of image processing K.Attakitmongcol,Grape leaf disease detection from
techniques and data mining algorithms, we were able to color imagery using hybrid intelligent system, 5th
locate the damaged leaf area in the plant. The numeric values International Conference on Electrical
of several image features are extracted. The algorithm Engineering/Electronics, Computer,
utilised in this scenario is extremely effective and achieves Telecommunications and Information Technology,
the lowest possible time-space complexity. The average Volume: 1,pp. 513 - 516,2008.
processing and clustering time for 90% of the photos is under 8. E.Omrani, B.Khoshnevisan, S.Shamshirband,
20 seconds. In every query image of the dataset, accuracy of H.Saboohi, N.B.Anuar, M.H.N.M.Nasir,'Potential of
at least 90% is attained. radial basis function-based support vector regression

Reference for apple disease detection', Department of

1. Evolutionary Artificial Neural Networks in Neutron Biosystem Engineering,pp.2-19,2014.

Spectrology, Jose Mauel OrtizRodriguez, Ma. Del 9. A.Singh, B.G.Subramanian, A.K.Singh, S.Sarkar,

Rosario Martinez-Blanco and Hector Rene Vega- Machine Learning for High- Throughput

Carrillo StressPhenotypinginPlants,TrendsinPlantScience,Vo

2. S.Bashir,N.Sharma,Remote Area Plant disease lume21,Issue2,pp.110- 124,2016.

detection using Image Processing, IOSR Journal of 10. A.Camargo, J.S.Smith, An image- processing based

Electronics and Communication Engineering , algorithm to automatically identify plant disease

Volume 2, Issue 6 ,pp.31 -34,2012. visual symptoms, Bio systems Engineering ,Volume

3. S.Arivazhagan,R.N.Shebiah, 102, Issue 1, pp. 9-21,2009.

S.Ananthi,S.V.Varthini ,Using texture detection 11. LeafDiseasedetectionofthecottonplantsusingimagepr

features, recoginizing unhealthy region of plant ocessingtechniques,Pranit Gulve, Sharayu S. Tambe,

leaves and classification of plant leaf disease , published in 2015

AgricEngInt: CIGR Journal, Vol. 15, No.1,pp.211- 12. An Investigation Into Machine Learning Regression

217,2013. Techniques for the Leaf Rust Disease Detection

4. J.Behmann,A.K.Mahlein,T.Rumpf,C.Romer,L.Plum Using Hyper spectral Measurement, IEEE Journal of

er, A review of advanced machine learning methods Selected Topics in Applied Earth Observations and

for the detection of biotic stress in precision crop Remote Sensing 9, 2016.

protection, Springer Media New Y ork, Precision 13. Chen CH, Huang WT, Tan TH, Chang CC, Chang

Agriculture, Volume 16, Issue 3, pp. 239–260, 2015. YJ, Using K nearest neighbour classification to

5. J.G.A.Barbedo, Digital image processing techniques detect abnormal Lung Sound, Sensors (base).

for detecting, quantifying and classifying plant 14. M.Jhuria, A.Kumar, R.Borse, Image processing for

diseases, ArnalBarbedo Springer Plus , pp.1- Smart Farming : Detection of disease and food

12,2013. grading, Institute of Electrical and Electronics

6. F.Qin, D.Liu, B.Sun, L.Ruan, Z.Ma, H.Wang, Engineering, 2013.

Identification of Alfalfa Leaf Diseases Using Image 15. N.Bhardwaj, G. Kaur, P.K.Singh, A Systematic

Recognition Technology, Plos Journals, pp.1-15, Review on Image Enhancement Techniques,

2016. Sensors and Image Processing, Advances in


Intelligent Systems and Computing (AISC)

You might also like