ISSN 2394-3777 (Print)

ISSN 2394-3785 (Online)

Available online at www.ijartet.com

International Journal of Advanced Research Trends in Engineering and Technology (IJARTET)

Vol. 7, Issue 4, April 2020

Insect Detection Using Image Processing and

IoT
Sanket Kasturiwala1, Smriti Banode2, Harsha Bhansali3, Riya Bhajan4, Aayushi Khandelwal5
Assistant Professor1, Undergraduate Students2,3,4,5, Electronics and Communications Engineering, Shri Ramdeobaba
College of Engineering and Management, Nagpur, India
Abstract: We propose a software technique to automatically identify the insect infesting the plant. We have particularly
chosen the region of Vidarbha (Maharashtra, India) and insects associated with the crops of this region. It is an
improvement in existing technology as it provides faster and more accurate solution. The developed scheme consists of 3
main phases. The first phase is training. The first stage is segmentation which uses saliency mapping, which is followed by
Feature extraction. In feature extraction we extract component values of colour map, texture map and edge map. Then the
extracted values are then fed to the database for further comparison. Training is followed by Testing where we give the
input to our program and test if it recognises the object (in our case, insect). And the third and the last phase includes the
introduction of server to upload images, hence making it accessible at any part of the world at all times.

Keywords: Insect Detection, Selective Rendering, Saliency Mapping, Support Vector Machine, Internet of Things, Image
Processing

I. INTRODUCTION III. PROPOSED METHODOLOGY AND WORKING

Plant diseases have caused major havoc as it can The general concept, which is the framework for any
cause substantial diminution in both quality and quantity image related algorithm of image classification is almost
of agricultural products. It causes massive damage. It is similar. First, the digital images in need of identification
estimated that in India, crop loss due to pests ranges from is uploaded on the server. Then image-processing
10-30 percent a year, out of which 26 % is due to the algorithms are used on the acquired images to extract
insect pests (Sriharan,1991). The changing agro-climatic informative features that are necessary for further
conditions contributed largely to the infestation of analysis. After that, several analytical discerning
various insect pests in different parts of the country. The techniques are used to categorize the images according to
naked eye observation of experts is the main approach the specific problem at hand.
adopted in practice for detection and identification of Fig 1: procedure of the proposed image processing-based disease
insects of plant. Automatic detection of insects is an detection solution
Image Acquisition
essential research topic as it may prove benefits in
monitoring large fields of crops.
In conclusion, the aim of this work is threefold: 1)
identifying the infested insect(s) using saliency mapping
procedure; 2) extracting the features set of the insects Image Pre-Processing
using colour co-occurrence methodology for texture
analysis; 3) detecting and classifying the type of insect
using SVM

II. RELATED WORK Image Segmentation

Work has been done for leaf disease detection as
mentioned in [1] but our project emphasises on detection
of insects that infest on the plant
Feature Extraction

All Rights Reserved © 2020 IJARTET 10

 ISSN 2394-3777 (Print)
ISSN 2394-3785 (Online)
Available online at www.ijartet.com

International Journal of Advanced Research Trends in Engineering and Technology (IJARTET)

Vol. II, Issue I, January 2015

Fig 2: Basic steps describing the proposed algorithm (Training)

Database Analysis

RGB image Acquisition

Classification using
Classifier
RGB image Acquisition Segmentation using Saliency
Mapping

Segmentation using Saliency Quaternion Domain

Mapping
Entropy
Quaternion Domain
Gaussian Filter is applied
Entropy
Taking Mean Values  Rough Salient Map
Gaussian Filter is applied
Indexing According to Values of mean
Taking Mean Values  Rough Salient Map
Image Super Imposition  Salient
Indexing According to Values of mean
Iamge

Image Super Imposition  Salient

Iamge Feature Extraction (Maps)

Texture
Colour

Feature Extraction (Maps) Edge

Colour Texture
Compare the extracted feature
Edge values with the existing
database and find the nearest
match (maximum resemblance)

Name of the insect identified

Save the Data in Database
All Rights Reserved © 2020 IJARTET 11
 ISSN 2394-3777 (Print)
ISSN 2394-3785 (Online)
Available online at www.ijartet.com

International Journal of Advanced Research Trends in Engineering and Technology (IJARTET)

Vol. II, Issue I, January 2015

phase SVM takes a data matrix as input data and labels

Fig 3: Basic steps describing the proposed algorithm (Testing)
each and every samples as either fitting to a given class
Saliency Map (positive) or not (negative). SVM considers each sample
A saliency Map shows unique quality of each pixel in the matrix as a row in a input space or high
though an image. Its goal is to represent an image in such dimensional feature space, where the number of traits
a way that it is informative and easier to analyse classifies the space dimensions. SVM learning procedure
construes the best hyperplane which divides each
Formulae Used in the above mentioned algorithm: positive and negative training sample. The trained SVM
can be used to predict about test samples.
Input Feature Maps
R = r-(g+b)/2;G = g-(r+b)/2;B = b-(r+g)/2;Y = (r+g)/2-
abs(r-g)/2-b;Y(Y<0) = 0;

Edge/texture map
Energy formulation
Overall energy
+

Naturally, if mi = mj then w mi,mj i,j equals zero

(texturing from the same view does not produce seams).
N denotes a set of pairs of adjacent faces, as in[8]
Fig. 4. Input Interface
Entropy

As in [5]
Correlation

Where i, j are the spatial coordinates of the function p (i, Fig. 5. Output Interface
j), Ng is grey tone, as in [5].
Support Vector Machine
SVM classification is based on the concept of
decision hyperplanes that ascertain decision boundaries
in input space or high dimensional feature space. SVM
creates linear functions from a set of categorized training
dataset. This hyperplane will try to separate the positive
samples from the
Negative samples. The linear separator is generally
structured with maximum distance from the hyperplane
to the closest negative and positive samples. This causes
precise classification for training data which is near, but Fig. 6. Server
not equal to the testing data. All through the training

All Rights Reserved © 2020 IJARTET 12

 ISSN 2394-3777 (Print)
ISSN 2394-3785 (Online)
Available online at www.ijartet.com

International Journal of Advanced Research Trends in Engineering and Technology (IJARTET)

Vol. II, Issue I, January 2015

IV. CONCLUSION [8]. “Seamless Mosaicing of Image-Based Texture Maps”, Victor

Lempitsky, Denis Ivanov, Department of Mathematics and
This project is made in such a way that it uses IoT in Mechanics ,Moscow State University
the form of server to make this product accessible to
anyone in any part of the world at any time. It is very
accurate and quick as compared to other solutions BIOGRAPHY
available out there [7] (apps name). It uses the latest
technology and algorithm such as SVM [6] and saliency Dr. Sanket Kasturiwala -Assistant
mapping .It is very handy as information is available to Professor, Electronics and
you with the click of a button. Communication Department, Shri
Ramdeobaba College of Engineering and
ACKNOWLEDGMENT Management.
Photograph
We would like to acknowledge Dr. Sanket
Kasturiwala (Assistant Professor, Electronics and
Communication Department, Shri Ramdeobaba College Miss. Smriti Banode pursuing her
of Engineering and Management) for being our Mentor Bachelor’s Degree in Electronics and
and Project Guide under whose guidance we were able to Communication Engineering from Shri
develop this technique of detecting insect and publish Ramdeobaba College of Engineering
this paper.Also we convey our sincere gratitude to Dr. and Management.
Panjabrao Deshmukh Krishi Vidyapeeth, College of
Agriculture (Nagpur) and Central Institute for Cotton
Research, whose assistance gave us the direction for the Miss. Harsha Bhansali pursuing her
project. Bachelor’s Degree in Electronics and
Communication Engineering from Shri
REFERENCES Ramdeobaba College of Engineering
[1]. H.Al-Hiary, S.Bani-Ahmad, M.Reyalat, M.Braik and Z. and Management.
ALRahamneh, ” Fast and Accurate Detection and Classification
of Plant Diseases,” International Journal of Computer
Applications (0975 – 8887), Volume 17– No.1, March 2011
Miss. Riya Bhajan pursuing her
[2]. Sikha O K1 , Sachin Kumar S2 , K P Soman2,” Salient Region
Detection and Segmentation in Images using Dynamic Mode
Bachelor’s Degree in Electronics and
Decomposition,” 1Department of Computer Science,2Centre for Communication Engineering from Shri
Computational Engineering and Networking,Amrita Viswa Ramdeobaba College of Engineering
Vidyapeetham, Coimbatore - 641 112 and Management.
[3]. Roland Br´emond and Josselin Petit and Jean-Philippe Tarel,”
Saliency Maps of High Dynamic Range Images,” Universi´e
Paris Est, LEPSiS, INRETS-LCPC
[4]. Anatomy of a color histogram - Computer Vision and Pattern
Recognition, 1992. Proceedings CVPR '92., 1992 IEEE Miss. Aayushi Khandelwal pursuing
Computer Society C
her Bachelor’s Degree in Electronics
[5]. Image Texture Feature Extraction Using GLCM Approach, P. and Communication Engineering
Mohanaiah*, P. Sathyanarayana**, L. GuruKumar***, from Shri Ramdeobaba College of
International Journal of Scientific and Research Publications,
Volume 3, Issue 5, May 2013
Engineering and Management.
[6]. Fereshteh Falah Chamasemani ; Yashwant Prasad Singh,” Multi-
class Support Vector Machine (SVM) Classifiers -- An
Application in Hypothyroid Detection and Classification,” 2011
Sixth International Conference on Bio-Inspired Computing:
Theories and Applications
[7]. PictureInsect- app on playstore

All Rights Reserved © 2020 IJARTET 13