CAMERA VISION BASED ANIMAL BEAT BACK SYSTEM FOR

AGRICULTURE USING MACHINE LEARNING

1
Mrs.Sivapriya K, 2Archana T, 3Jayanandhini S, 4Jayasri G, 5Keerthana K, 6Keerthika.S
1
Assistant professor, Department of Computer Science and Engineering, Vivekanandha College of
Engineering for Women, Tiruchengode, Namakkal District, TamilNadu, 637205.
2,3,4,5,6
Student, Department of Computer Science and Engineering, Vivekanandha College of
Engineering for Women, Tiruchengode, Namakkal District, TamilNadu, 637205.

ABSTRACT: Most farmers have challenges related to crop

As a result of human interference with damage due to wildlife pests. Animal intrusion
natural habitats and deforestation, crop is a major threat to the productivity of the crops,
raiding by animals has emerged as one of the which affects food security and reduces the
most prevalent human-animal conflicts. Wild profit to the farmers. Organic farmers have
animals can assault farmers working in the additional challenges because they cannot use
fields and seriously harm agricultural chemical controls which are sometimes the
harvests. Due to agricultural raiding by wild most effective and efficient options. A need has
animals like elephants, wild boar, and deer, been identified for alternative pest control
farmers experience significant crop loss. The appropriate for traditional and organic farmers.
protection of crops against assaults by wild Three types of animal intrusion you might find
animals is one of the primary concerns of include animal tracks, crop damage and animal
today's farmers. There are numerous scat or faeces. In the case of animal tracks, only
conventional methods to deal with this issue, one instance of tracks in the field carries a
both lethal (such as shooting and trapping) relatively low risk.
and non- lethal. (e.g., scarecrow, chemical 2. LITERATURE SURVEY
repellents, organic substances, mesh, or Zachar J.Ruff, Damon B.Lesmeister, Cara
electric fences). The edge computing device L.Appel, Christopher M.Sullivan
turns on the camera, then uses its DCNN 2021 “Workflow and
software to identify the target. If an animal is convolutional neural network for automated
discovered, it then sends a message to the identification of animal sounds” The aim of this
Animal Repelling Module with information project is automated identification of animal
about the type of ultrasound that should be sounds using Convolutional Neural
created based on the animal's category. Networks.[1]R. Nikhil;
Keyword: Animal Recognition, Repellent, B.S. Anisha; RamakanthKumar.P2020.“Real-
Artificial Intelligence, Edge Computing, Time Monitoring of Agricultural Land with
Animal Detection, Deep Learning, DCNN. Crop Prediction and Animal Intrusion
1. INTRODUCTION: Prevention using Internet of Things and
Agriculture has seen many revolutions, whether Machine Learning at Edge” The aim of this
the domestication of animals and plants a few project is crop prediction helps the farmers to
thousand years ago, the systematic use of crop grow suitable crops depending on the soil
rotations and other improvements in farming parameters by the use of machine learning
practice a few hundred years ago, or the “green techniques and it also helps in prevention of the
revolution” with systematic breeding and the intruders like wild animals into the field. [2] L.
widespread use of man-made fertilizers and G. C. Vithakshana; W. G. D. M. Samankula
pesticides a few decades ago. “IoT based animal classification system
using convolutional neural
network”. The aim of this project is an IoT

ISSN (PRINT): 2393-8374, (ONLINE): 2394-0697, VOLUME-10, ISSUE-4, 2023

DOI: 10.21276/ijcesr.2023.10.4.21
196
 INTERNATIONAL JOURNAL OF CURRENT ENGINEERING AND SCIENTIFIC RESEARCH (IJCESR)
based acoustic classification system was other types. Additionally, these fences are
designed using Convolutional Neural Networks widely acceptable and meet various regulations.
(CNN), which is beneficial for those who are Their disadvantage includes their short lifespan
interested in monitoring ecosystems such as (up to 10 years) and questionable effectiveness
animal scientists, zoologists, and in areas with a higher possibility of wild animal
environmentalists..[3] Henry Roberts; Aviv crop damage.
Segev 2020. “Animal Behaviour Prediction
with Long Short-Term Memory”. The aim of
this project is efficiently converting video of
animals at any length into models capable of
making accurate behavioural prediction using
Long Short-term Memory (LSTM).[4] Loris
Nanni , Sheryl Brahnam , Alessandra Lumini
and Gianluca Maguolo 2020. “Animal Sound Fig:3.1 wire and plastic fences
Classification Using Dissimilarity Spaces”. The
aim of this project is to automated animal audio • Electric fences
classification using clustering methods..[5] S. These are constructed to inflict an electric shock
Jeevitha; S.Vengatesh Kumar 2019. “A Study to animals that come in contact with the fence,
on Sensor Based Animal Intrusion Alert System thus preventing animals from crossing the
Using Image Processing Techniques”. an fence. These fences are long lasting and an
animal intrusion alert system is designed by effective crop protection measure. Costs vary
employing wireless sensors for sending an depending on specific type and size of an area.
automatic alert message to both the landowner Before purchasing electric fences, it's very
and forest officials with an image.[6] important to make sure they are allowed for use
3. EXIXTING SYSTEM in the specific area, and for protection against
Wild animals are a special challenge for farmers endangered animal species. Additionally, it's
throughout the world. Animals such as deer, recommended that electric fences are marked
wild boars, rabbits, moles, elephants, monkeys, with a warning sign to prevent any possible
and many others may cause serious damage to human contact.
crops. They can damage the plants by feeding Its disadvantages include the potential for the
on plant parts or simply by running over the entire fence to be disabled due to a break in the
field and trampling over the crops. Therefore, conducting wire, shorting out if the conducting
wild animals may easily cause significant yield wire contacts any non-electrified component
losses and provoke additional financial that may make up the rest of the fence, power
problems. Another aspect to consider is that failure.
wild animal crop protection requires a 4. PROPOSED SYSTEM
particularly cautious approach. In other words, AI Computer Vision based DCNN for detecting
while utilizing his crop production, every animal species, and specific ultrasound
farmer should be aware and take into emission (i.e., different for each species) for
consideration the fact that animals are living repelling them.design, deployment and
beings and need to be protected from any assessment of an intelligent smart agriculture
potential suffering. repelling and monitoring IoT system based on
embedded edge AI, to detect and recognize the
• Wire fences different kinds of animal, as well as generate
constructed of metal wires woven together ultrasonic signals tailored to each species of the
forming a physical barrier. The fences are animal.This combined technology used can help
effective, long lasting, and require relatively farmers and agronomists in their decision
little maintenance. However, they are expensive making and management process.
and recommended only for the protection of Deep learning in the form of Convolutional
high- value crops. Neural Networks (CNNs) to perform the animal
• Plastic fences recognition.
polypropylene fences are generally less
expensive and easier to install and repair than

ISSN (PRINT): 2393-8374, (ONLINE): 2394-0697, VOLUME-10, ISSUE-4, 2023

DOI: 10.21276/ijcesr.2023.10.4.21
197
 INTERNATIONAL JOURNAL OF CURRENT ENGINEERING AND SCIENTIFIC RESEARCH (IJCESR)
• DCNN • Disk space: 320 GB
CNNs are a category of Neural Networks that • Operating systems: Windows® 10,
have proven very effective in areas such as macOS*, and Linux*
image recognition and classification. CNNs are SOFTWARE REQUIREMENTS
a type of feed-forward neural networks made up • Server Side : Python 3.7.4(64-bit) or
of many layers. CNNs consist of filters or (32-bit)
kernels or neurons that have learnable weights • Client Side : HTML, CSS, Bootstrap
or parameters and biases. Each filter takes some • IDE : Flask 1.1.1
inputs, performs convolution and optionally • Back end : MySQL 5.
follows it with a non-linearity. A typical CNN • Server : Wampserver 2i
architecture can be seen as shown in Fig.3.1. • OS : Windows 10 64 –bit or Ubuntu
The structure of CNN contains Convolutional, 18.04 LTS “Bionic Beaver”
pooling, Rectified Linear Unit (ReLU), and 5.2.1.SOFTWARE
Fully Connected layers. Python 3.7.4
Python is a general-purpose interpreted,
interactive, object- oriented, and high-level
programming language. It was created by Guido
van Rossum during 1985- 1990. Like Perl,
Python source code is also available under the
GNU General Public License (GPL). This
Fig:4.1.CNN tutorial gives enough understanding on Python
programming language.
The detection system recorded the date and time
of each detection. In addition, there were
cameras and a video recording system that
recorded all animal movements within the
enclosure. The detection log was compared to
the images from the cameras, which also had a
date and time stamp, to investigate the Python is a high-level, interpreted, interactive
reliability of the system. A message alert is sent and object- oriented scripting language. Python
to the registered mobile number. is designed to be highly readable. It uses
Animals generally have a sound sensitive English keywords frequently where as other
threshold that is far higher than humans. They languages use punctuation, and it has fewer
can hear sounds having lower frequencies with syntactical constructions than other languages.
respect to the human ear. For instance, while the Python is a MUST for students and working
audible range for humans is from 64Hz professionals to become a great Software
- 23KHz, the corresponding range of goats, Engineer specially when they are working in
sheep, domestic pigs, dogs and cats is 78Hz - Web Development Domain.
37KHz, 10Hz - 30KHz, 42Hz - 40.5KHz 67Hz
- 45KHz and 45Hz - TENDER FLOW
64KHz.Generating ultrasounds within the TensorFlow is an end-to-end open-source
critical perceptible range causes animals to be platform for machine learning. It has a
disturbed, thus making them move away from comprehensive, flexible ecosystem of tools,
the sound source. At the same time, these libraries, and community resources that lets
ultrasounds are not problems to the human ear researchers push the state-of-the-art in ML, and
even when the frequency range is beyond the gives developers the ability to easily build and
human. deploy ML-powered applications.TensorFlow
5. SYSTEM REQUIREMENTS provides a collection of workflows with
HARDWARE REQUIREMENTS intuitive, high-level APIs for both beginners
• Processors: Intel® Core™ i5 processor and experts to create machine learning models
4300M at in numerous languages. Developers have the
2.60 GHz or 2.59 GHz (1 socket, 2 cores, 2 option to deploy models on a number of
threads per core), 8 GB of DRAM platforms such as on servers, in the cloud, on

ISSN (PRINT): 2393-8374, (ONLINE): 2394-0697, VOLUME-10, ISSUE-4, 2023

DOI: 10.21276/ijcesr.2023.10.4.21
198
 INTERNATIONAL JOURNAL OF CURRENT ENGINEERING AND SCIENTIFIC RESEARCH (IJCESR)
mobile and edge devices, in browsers, and on Matplotlib
many other JavaScript platforms. This enables Matplotlib is a comprehensive library for
developers to go from model building and creating static, animated, and interactive
training to deployment much more easily. visualizations in Python. Matplotlib makes easy
things easy and hard things possible

Matplotlib is a plotting library for the Python

programming language and its numerical
mathematics extension NumPy. It provides an
object-oriented API for embedding plots into
applications using general-purpose GUI toolkits
Keras like Tkinter, wxPython, Qt, or GTK.
Keras is a deep learning API written in Python,
running on top of the machine learning platform
Tensor Flow. It was developed with a focus on
enabling fast experimentation.
.
MySQL
MySQL tutorial provides basic and advanced
concepts of MySQL. Our MySQL tutorial is
designed for beginners and professionals.
MySQL is a relational database management
Allows the same code to run on CPU or on system based on the Structured Query
GPU, seamlessly. Language, which is the popular language for
User-friendly API which makes it easy to accessing and managing the records in the
quickly prototype deep learning models. database. MySQL is open-source and free
Built-in support for convolutional networks (for software under the GNU license. It is supported
computer vision), recurrent networks (for by Oracle Company.MySQL database that
sequence processing), and any combination of provides for how to manage database and to
both. manipulate data with the help of various SQL
Supports arbitrary network architectures: multi- queries. These queries are: insert records,
input or multi-output models, layer sharing, update records, delete records, select records,
model sharing, etc. This means that Keras is create tables, drop tables, etc. There are also
appropriate for building essentially any deep given MySQL interview questions to help you
learning model, from a memory network to a better understand the MySQL database
neural Turing machine.
PANDAS
pandas is a fast, powerful, flexible and easy to
use open source data analysis and manipulation
tool, built on top of the Python programming
language. pandas is a Python package that
provides fast, flexible, and expressive data MySQL is currently the most popular database
structures designed to make working with management system software used for
"relational" or "labeled" data both easy and managing the relational database.
intuitive. It aims to be the fundamental high-
level building block for doing practical, real
world data analysis in Python.

NumPy is a general-purpose array-processing

package. It provides a high-performance
multidimensional array object, and tools for
working with these arrays.

ISSN (PRINT): 2393-8374, (ONLINE): 2394-0697, VOLUME-10, ISSUE-4, 2023

DOI: 10.21276/ijcesr.2023.10.4.21
199
 INTERNATIONAL JOURNAL OF CURRENT ENGINEERING AND SCIENTIFIC RESEARCH (IJCESR)
Wamp Server Flask is a web framework. This means flask
WampServer is a Windows web development provides you with tools, libraries and
environment. It allows you to create web technologies that allow you to build a web
applications application.

This web application can be some web pages, a

blog, a wiki or go as big as a web-based
calendar application or a commercial website.

Flask is often referred to as a micro framework.

It aims to keep the core of an application simple
yet extensible. Flask does not have built-in
abstraction layer for database handling, nor
WAMPServer is a reliable web development does it have formed a validation support.
software program that lets you create web apps Instead, Flask supports the extensions to add
with MYSQL database and PHP Apache2. With such functionality to the application. Although
an intuitive interface, the application features Flask is rather young compared to most Python
numerous functionalities and makes it the frameworks, it holds a great promise and has
preferred choice of developers from around the already gained popularity among Python web
world. The software is free to use and doesn’t developers. Let’s take a closer look into Flask,
require a payment or subscription. so-called “micro” framework for Python.
Bootstrap 4
Bootstrap is a free and open-source tool 6. SYSTEM ARCHITECTURE
collection for creating responsive websites and
web applications. It is the most popular HTML,
CSS, and JavaScript framework for developing
responsive, mobile-first websites.

Animal

Activate Alert Intrude

It solves many problems which we had once,
one of which is the cross-browser compatibility Update Animal

issue. Nowadays, the websites are perfect for all

the browsers (IE, Firefox, and Chrome) and for
all sizes of screens (Desktop, Tablets, Phablets,
and Phones). All thanks to Bootstrap developers
-Mark Otto and Jacob Thornton of Twitter,
though it was later declared to be an open-
source project.

Fig.6.1 System Architecture

Flask

ISSN (PRINT): 2393-8374, (ONLINE): 2394-0697, VOLUME-10, ISSUE-4, 2023

DOI: 10.21276/ijcesr.2023.10.4.21
200
 INTERNATIONAL JOURNAL OF CURRENT ENGINEERING AND SCIENTIFIC RESEARCH (IJCESR)
ANIMAL-10N dataset contains 5 pairs of
6.1.DCNN Model confusing animals with a total of 55,000
images. The 5 pairs are as following: (cat,
Test
lynx), (jaguar, cheetah), (wolf, coyote),
(chimpanzee, orangutan), (hamster, guinea pig).
The images are crawled from several online
Animal search engines including Bing and Google using
DL Model DL Model the predefined labels as the search keyword.
Preprocessing Capture Live Pre-processing
Segmentation Preprocessing Animal Image pre-processing are the steps
Feature Segmentation
taken to format images before they are used by
model training and inference. The steps to be
Classification Feature taken are:
• Read image
• RGB to Grey Scale conversion
• Resize image
Original size (360, 480, 3) — (width, height,
no. RGB channels)
Resized (220, 220, 3)
Animal Not Matching Animal Found • Remove noise (Denoise)
Animal Type
ANIMAL CLASSIFICATION
Repellent Alert Farmer
DCNN algorithms were implemented to
automatically detect and reject improper animal
images during the classification process. This
7. MODULES will ensure proper training and therefore the
• Animal Repellent Web Dashboard best possible performance.
• Animal Recognition
• Animal Identification
• Repellent
• Monitoring and Visualizing
• Notification
• Performance Analysis
7.1.ANIMAL REPELLEANT WEB
DASHBOARD
This system is economical as compared to many
of the existing solutions like electric fences, ANIMAL IDENTIFICATION
brick walls and manual supervision of the After capturing the animal image from the Farm
fields.This system is very effective in driving Camera, the image is given to animal detection
off the animals from the fields and keeping module. This module detects the image regions
them away. It accurately determines the which are likely to be human. After the animal
presence of animals in the fields and sounds the detection using Region Proposal Network
buzzer. It does not sound the buzzer due to the (RPN), animal image is given as input to the
presence of a human being or due to some feature extraction module to find the key
random motion features that will be used for classification. The
ANIMAL RECOGNITION module composes a very short feature vector
Training Phase that is well enough to represent the animal
This module begins by annotation of animal image.
dataset. These templates then become the Prediction
reference for evaluating and registering the In this module the matching process is done
templates for the other poses: tilting up/down, with trained classified result and test animal
moving closer/further, and turning left/right. image with animal dataset classified file.
Animal Image Acquisition Hamming Distance is used to calculate the

ISSN (PRINT): 2393-8374, (ONLINE): 2394-0697, VOLUME-10, ISSUE-4, 2023

DOI: 10.21276/ijcesr.2023.10.4.21
201
 INTERNATIONAL JOURNAL OF CURRENT ENGINEERING AND SCIENTIFIC RESEARCH (IJCESR)
difference according to the result the prediction and accuracy are calculated using the True
accuracy will be displayed. Here, it is done with positive (TP), true negative (TN), false negative
DCNN with the help of a pattern classifier, the (FN), and false positive (FP). TP is the number
extracted features of animal image are of positive cases that are classified as positive.
compared with the ones stored in the animal FP is the number of negative cases that are
database. The animal image is then classified as classified as positive. TN is the number of
animal type. If the animal is found negative cases classified as negative and FN is
corresponding repellent module is called. the number of positive cases classified as
negative
REPELLENT
Monitoring window detecting the presence of
animals then it enables repeller module to
repelling them through the generation of
ultrasounds, which has recently been proven as
an alternative, effective method for protecting
crops against predicted animals. Animals
Accuracy
generally have a sound sensitive threshold that
Accuracy is a measure that tells whether a
is far higher than humans. They can hear sounds
model/algorithm is being trained correctly and
having lower frequencies with respect to the
how it performs. In the context of this thesis,
human ear. For instance, while the audible
accuracy tells how well it is performing in
range for humans is from 64Hz - 23KHz, the
detecting humans in underwater environment.
corresponding range of goats, sheep, domestic
Accuracy is calculated using the following
pigs, dogs and cats is 78Hz - 37KHz, 10Hz -
formula.
30KHz, 42Hz - 40.5KHz 67Hz - 45KHz and
Accuracy = (T P + T N)/ (T P + T N + F P + F
45Hz - 64KHz
N)
respectively.
8. RESULT AND DISCUSSION
Monitoring and Visualizing
The system works in real time detect the
Fig 8.1. Training and Validation accuracy and
animals in the field, in addition the farmers can
access the view of their fields remotely. Type of
animal and also the count can be given. The
animal recognition module will share the data
over the cloud regularly through a Wi-Fi
connection. The cloud setup will consist of a
private cloud instance running on a machine.
The data shared will be used to analyse the
patterns and responses of wild animals. The
farmer can visualize the errors if any, resolve
them, and achieve better results.
Notification loss graph
The email and sms notification consisting of
captured image is notified to the user regarding
the detected motion in this phase. The email is
sent to registered email id and sms is sent to the
Telegram account of the user to the registered
number.
Performance Analysis
In this module we able to find the performance Fig.8.2. (a) accuracy of the joint CNN (Top-1
of our system using SENSITIVITY, and Top-5) during training; (b) training and
SPECIFICITY AND ACCURACY of testing accuracy of the joint CNN (Top-5).
Data in the datasets are divided into two classes
not animal (the negative class) and animal and
type (the positive class). Sensitivity, specificity,

ISSN (PRINT): 2393-8374, (ONLINE): 2394-0697, VOLUME-10, ISSUE-4, 2023

DOI: 10.21276/ijcesr.2023.10.4.21
202
 INTERNATIONAL JOURNAL OF CURRENT ENGINEERING AND SCIENTIFIC RESEARCH (IJCESR)
Table 8.3. Animal recognition results using a
single branch SVM.
AP, AP,
% MR,% FP,% Animals % FP,%
78.83 12.2 11.3 Goat 76.52 16.3
80.64 9.4 14.8 Cow 81.3 12.5
73.88 15.1 18.6 Elephant 75.54 17.9
76.74 13.7 17.4 Deer 69.88 22.4
80.91 8.9 16.1 Horse 63.85 26.9
79.72 8.1 18.2 Pig 79.21 18.7
Table 8.4. Animal recognition results using the
joint CNN.
AP, AP,
Fig.8.8.Preprocessing
% MR,% FP,% Animals % MR,% FP,%
83.37 10.7 9.7 Goat 81.93 6.9 14.4
84.29 9.1 13.2 Cow 85.69 7.1 10.8
79.13 12.4 16.3 Elephant 79.14 12.3 14.7
80.21 11.8 16.5 Deer 75.35 18.6 19.8
84.9 7.9 12.9 Horse 69.52 21.6 24.8
83.07 8 15.6 Pig 81.23 14.9 17.2

Fig.8.9.Feature Selection

Fig.8.5.Animal monitoring

Fig.8.10.farmer Information

Fig.8.6.Register process

Fig.8.7.Training process Fig.11.Animal Detection Information

ISSN (PRINT): 2393-8374, (ONLINE): 2394-0697, VOLUME-10, ISSUE-4, 2023

DOI: 10.21276/ijcesr.2023.10.4.21
203
 INTERNATIONAL JOURNAL OF CURRENT ENGINEERING AND SCIENTIFIC RESEARCH (IJCESR)
9. CONCLUSION [3] M. S. Farooq, S. Riaz, A. Abid, K. Abid,
Agricultural farm security is widely needed and M. A. Naeem, ``A surveyon the role of IoT
technology nowadays. In order to accomplish in agriculture for the implementation of smart
this, a vision-based system is proposed and farming,''IEEE Access, vol. 7, pp. 156237-
implemented using Python and Open CV and 156271, 2019.
developed an Animal Repellent System to blow [4] K. Kirkpatrick, ``Technologizing
out the animals.The implementationof the agriculture,'' Commun. ACM, vol. 62,no. 2, pp.
application required the design and 14-16, Jan. 2019.
development of acomplex system for intelligent [5] A. Farooq, J. Hu, and X. Jia, ``Analysis
animal repulsion, which integratesnewly of spectral bands and spatialresolutions for
developed software components and allowsto weed classification via deep convolutional
recognize the presence and species of animals neural network,''IEEE Geosci. Remote Sens.
in realtime and also to avoid crop damages Lett., vol. 16, no. 2, pp. 183-187,Feb. 2018.
caused by the animals.Based on thecategory of [6] M. Apollonio, S. Ciuti, L. Pedrotti, and
the animal detected, the edge computing P. Banti,
deviceexecutes its DCNN Animal Recognition ``Ungulates and theirmanagement in Italy,'' in
model to identify the target, and if ananimal is European Ungulates and Their Managementin
detected, it sends back a message to the the 21th Century. Cambridge, U.K.: Cambridge
AnimalRepelling Module including the type of Univ. Press, 2010,pp. 475- 505.
ultrasound to be generatedaccording to the [7] A. Amici, F. Serrani, C. M. Rossi, and
category of the animal. The proposed CNN was R. Primi,
evaluatedon the created animal database. The ``Increase in crop damagecaused by wild boar
overallperformances were obtained using (Sus scrofa L.): The `refuge effect,'''
different number oftraining images and test AgronomySustain. Develop., vol. 32, no. 3, pp.
images.The obtained experimental resultsof the 683-692, Jul. 2012.
performed experiments show that the [8] S. Giordano, I. Seitanidis, M. Ojo, D.
proposedCNN gives the best recognition rate Adami, and F. Vignoli, ``IoT solutionsfor crop
for a greater numberof input training images protection against wild animal attacks,'' in Proc.
(accuracy of about 98 %). This project IEEE Int.Conf. Environ. Eng. (EE), Mar. 2018,
presented a real-time monitoring solution based pp. 1-5.
on AI technology to address the problems of [9] M. O. Ojo, D. Adami, and S. Giordano,
crop damages against animals.This technology ``Network performance evaluationof a LoRa-
used canhelp farmers and agronomists in their based IoT system for crop protection against
decision makingand management process. ungulates,'' inProc. IEEE 25th Int.Workshop
Comput. Aided Modeling Design
10. FUTURE ENHANCEMENT Commun.Links Netw. (CAMAD), Sep. 2020,
Further in the proposed architecture, some pp. 1-6.
image compression techniques can be [10] H. E. Heefner and R. S. Heffner,
developed to reduce the time taken for ``Auditory perception,'' in Farm Animalsand the
notification to reach user as described above. Environment, C. Phillips and D. Piggins, Eds.
Wallingford, U.K.:CAB International, 1992.
11. REFERENCES [11]https://www.indiatimes.com/news/india/in-
[1] M. De Clercq, A. Vats, and A. Biel, the-battle- between-man-vs-wild-highways-
``Agriculture 4.0: The future of railway- lines-emerge-as- new-challenges-
farmingtechnology,'' in Proc. World 375504.html
Government Summit, Dubai, UAE, 2018,pp. [12]https://adventuresinmachinelearning.com/n
11-13. eural- networks-tutorial/
[2] Y. Liu, X. Ma, L. Shu, G. P. Hancke, [13] https://github.com/tensorflow/models
and A. M. Abu- Mahfouz, ``Fromindustry 4.0 to
agriculture 4.0: Current status, enabling
technologies,and research challenges,'' IEEE
Trans. Ind. Informat., vol. 17, no. 6,pp.
4324334, Jun. 2021.

ISSN (PRINT): 2393-8374, (ONLINE): 2394-0697, VOLUME-10, ISSUE-4, 2023

DOI: 10.21276/ijcesr.2023.10.4.21
204