IRJET - IoT based Facial Recognition Quadcopter using Machine Learning Algorithm

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 07 Issue: 02 | Feb 2020 www.irjet.net p-ISSN: 2395-0072
© 2020, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 2640
IoT Based Facial Recognition Quadcopter Using Machine Learning
Algorithm
Shahriar Mostufa1
1UG Student, Dept. Electrical and Electronic Engineering, Rajshahi University of Engineering and Technology
(RUET), Rajshahi, Bangladesh
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract – In this paper, a block diagram and algorithmic
model is proposed on the basis of facial recognition and
machine learning algorithm. The effectivity of facial
recognition on Quadcopter is also shown in this paper. IoT is
nowadays a very common wordandanythingcanbecontrolled
wirelessly from any global distance. Here the quadcopter is
controlled using the internet of things so it is easy and flexible
of range of control area. The implementation of the machine
learning algorithm can make it faster and the processing
efficiently of huge data for machine learning is also given in
this paper.
Key Words: Internet of Thing (IOT), Raspberry Pi, facial
recognition, machine learning algorithm, Quadcopter,
Esp8266 etc.
1. INTRODUCTION
The IoT is now the most common term in today'sworld.Here
using this, the quadcopter can easily be controlled wirelessly
the esp8266 is used for this, again raspberry pi has a build-in
wireless connection module in it. Facial recognition
algorithm needs huge processing power and applying
machine learning algorithm on it needs a huge processor so
the data is showed how it is sent to the server and the
processing again the logical output for performing certain
tasks.
1.1 EFFECTIVITY
The main purpose of this is that as it is IoT based controlled
so it can freely move wherever it needed. Having a camera
with facial recognition algorithm is mostly benefited is
because it can move such a height and distance freely, it can
also track any people, object easily where the normal camera
is fixed in a particular position. Moreover, this can avoid
obstacles where a normal camera is failed to avoid. The
proposed model shows all the ins and outs of this
quadcopter.
2. PROPOSED SYSTEM
The following figure shows the architecture of the project
and its implementation.
Block Diagram of Proposed System
3. WORKING METHODOLOGY
 In this proposed system the main processing power
is raspberry pi. It is a very cheap computer that runs
Linux and it also has GPIO (general purpose
input/output) pins which allow controlling
components. Here the quadcopter is controlled
through these pins and all those logical instructions
given from admin will execute by this.
 Raspberry Pi is Linux supported so it can easily run
C, C++.Python language. Here Python is most
efficiently used for this system. Because Python has
many built-in machine learning and facial
recognition algorithms. The most common facial
recognition algorithm in python is Open CV.
Therefore, using this library in python, the
quadcopter can easily gain facial recognition ability.
There is also a library in Python called NumPy. By

using this library, the image data is slicedintoarrays
and the images are nothing but 2D of arrays. So, all
of these would be possible through Python.
 Raspberry Pi has a built-in Wi-Fi module which will
help us to gain IoT control. Esp8266 can easily
connect to Wi-Fi or the internet, so the quadcopter
can be controlled from anywhere in the world. This
module will be also used to transmit and receive
data from the server. As images are a huge amount
of data and the machine learning algorithm needs
more processing power so in order to enable thefull
capability of the quadcopter the data needs to send
to the server and process there.
 Control panel can easilycontrol thequadcopterfrom
the help of server or directly to the raspberry pi if it
is in a certain range of the Wi-Fi module ESP8266.
4. COMPONENTS AND SOFTWARE
4.1 HARDWARE
a. Raspberry Pi:
The Raspberry Pi is a small pocket-size computer
used to do small computing and networking
operations. It is the main element in the field of the
internet of things. It provides access to the internet
and hence the connectionof automationsystemwith
remote locationcontrollingdevice becomespossible.
Raspberry Pi is available in various versions. Here,
model Pi 2 model B is used and it has quad-core
ARM Cortex-A53 CPU of 900 MHz, and RAM of 1GB.
it also has 40 GPIO pins, Full HDMI port, 4 USB ports,
Ethernet port, 3.5mm audio jack, video Camera
interface (CSI), theDisplayinterface(DSI),andMicro
SD card slot.[1]
b. ESP8266 Module:
The ESP8266 is a low-cost Wi-Fi module consists of
a Wi-Fi chip with full TCP/IP stack and
microcontrollerchipmanufactured byM/SEspruino
[2]. ESP8266 uses serial transceiver (Tx/Rx)tosend
and receive data in Ethernet buffers, and serial
commands to query and change configurations of
the Wi-Fi module. It onlyrequirestwowires(Tx/Rx)
to communicate between a microcontroller and a
Wi-Fi module. It offloads Wi-Fi-related tasks to the
module, allowing the microcontroller code to be
very light-weighted. Wi-Fi Module is addressable
over SPI and UART, making it easy to build an
Internet of Thingsapplication.WeuseATcommands
to connect to Wi-Fi networks and open TCP
connections without the need to have TCP/IP stack
running in our microcontroller. By just directing
connecting the microcontroller to this module, we
can start pushing data up to the Internet (Central
server). [3]
SOFTWARE:
a. PHYTHON:
There are a large number of reasonsfor pythontobe
considered as a popular language for machine
learning. As stated by Harrington (2012) Pythonisa
simple and elegant language with clear syntax
having extremely easy text manipulation and
processing capabilities and a rich community of
developers leading to the availability of ample
documentation. Python also comes with an
interactive shell that allows the programmers to
examine and view elements of a program
simultaneously while writing the code. Python has
been equipped with high-level data types like lists,
queues, tuples, dictionaries, etc. that make it easy to
implement abstract concepts and do not need to be
programmed explicitly bytheprogrammer.Machine
learning could also be easily implemented using
other high-level languages like MATLAB that
provides a number of built-in features for
performing matrix mathematics but it is not an
open-source language. Moving towards languages
like C and Java, matrix math libraries are available
for them as well but it requires a lot of code to
perform simple things in these languages. These
limitations have been overcome by python thatisan
open-source, clean, brief and easy to read language
with the availability of a large number of libraries
and packages that make python more popular in
various domains including scientific and financial
communities. Moreover, python does not require
expert programming skills to catch up with its code,
even non-programmers could easily learn and code
in python.[8]
According to official python documentation, Scikit-
Learn is a simple, efficient and powerful tool that
provides various machine learning algorithms for
classification,clustering,regression,model selection,
etc. It is an open-source library easily accessible to
anybody and usable in many contexts. To work with
Scikit-Learn, NumPy must be installed along with
SciPy and Matplotlib. NumPy is used to work with
arrays; Matplotlib is used to plot graphs, Keras for
Preprocessing image, NLTK for wordnet, Pandas for
reading CSV data file, Pyttsx for text to speech, Text
blob for sentiment analysis, chatterbot for chatting
application and PIL for image manipulation.
Machine learning has improved the quality of life of
humans by providing several applications to
facilitate human living. Among the numerous

applications of machine learning in the field of
health, science, industries, etc.isthetimelydetection
of diseases such as cancer, glaucoma and other
diseases which are claiming human lives at a jaw-
breaking rate, the visualization of smart cars,
effective web search which has made the internet
searches easier, language translations are
immensely helping in worldwide communications
and limiting the great language barrier among
countries, realization of fraud detection and face
recognition systems to mentionbuta fewaregreatly
helping to improve the quality of life of humans. It is
in this regard that Machine Learning has remained
significant over the years.[7]
b. OPEN CV:
OpenCV stands for Open Source Computer Vision
Library and is designed in C & C++ specifically for
increased computational efficiency and supported
by most Operating Systems [4]. Example
applications of the OpenCV library include Human-
Computer Interaction (HCI), Object Identification,
Segmentation and Recognition, Face Recognition,
Gesture Recognition, Camera and Motion Tracking,
Ego Motion, Motion Understanding,StructureFrom-
Motion (SFM), Stereo and Multi-Camera Calibration
and Depth Computation and Mobile Robotics [4].
The basic goal of the Computer Vision application is
to make useful decisions about real objects in the
scene.
Understanding any activity involves being able to
detect and classify targets of interest and analyze
what they are doing. In this PG project, a student
demonstrates a process fordetectingmovingtargets
and extracting boundaries.Themainpurposebehind
the project is to implement robust method to detect
human if present in video under dynamic
environment. This can be used as an efficient visual
surveillance method. The software will monitor for
movement or motion in a localized area. If motion is
detected it will check for object whether it is human
or nonhuman. We collected a total of indoor and
outdoor video sequences captured from stationary
cameras. The method implemented gives very good
performance and detects human about 90% of time
[5].
OVERVIEW OF THE UAV(Unmanned AerialVehicle)
CHANNELS OF INTERACTION WITH THE PUBLIC
COMMUNICATIONS NETWORK :
One or more UAV can be used for the tasks of data collection
and transmission (figure 1). Autonomous UAV flight is
performed on the basis of the flight task, which is pre-loaded
into the controller (Kirichek et al. 2015). Although the full
autonomy of the flight, the telemetrychannel istypicallyused
for monitoring parameters of UAV and for supplying the
emergency landing command. Given the fact that the
maintenance traffic is transmitted in the channel – the
channel utilization is about 60%, the channel resource
remains small for efficient data transmission collected from
WBAN.[9]
figure: Data Collection from WBAN Using One or Multiple
UAVs
This shows how quadcopter can collect data and transmit it
to server over a channel and get command and logical
decision from the server.
MACHINE LEARNING ON SERVER:
Microsoft Machine Learning Server9.4includesspecializedR
packages and Python modules for developing and
operationalizing solutions written in R and Python. Machine
Learning Server 9.4 comes with version 3.5.2 of the open
source R language engine and a run-time infrastructureforR
script execution. On the Python side, it comes with an open
source distribution of Python and run-timeinfrastructurefor
script execution. Machine Learning Server 9.4 alsoincludesa
rich set of highly-scalable algorithms such as RevoScaleR,
revoscalepy, and Microsoft ML that can work on data sizes
larger than the size of physical memory, and run on a wide
variety of platforms in a distributed manner. Machine
Learning Server bridges these Microsoft innovations and
those coming from the open source community (R, Python
and AI toolkits) all on top of a single enterprise-grade
platform. Any R or Python open source machine learning
package can work side by side with any proprietary
innovation from Microsoft. MachineLearningServerrunson-
premises and in the cloud on a variety of platforms, in a
variety of deployments, and can support the need to have
machine learning and analytics across multiple platforms,
including clustered topologiesforApacheSpark™onHadoop.
On two data platforms, SQL Server and Apache Spark (on
HDFS), the R and Python libraries from Microsoft can
compute and analyze data locally, returning just the results,
without the need to pull data across the network. This
capability is called remote compute context. It requires theR

and Python libraries and interpreters fromMicrosofton both
client and server systems, but the client versions are free of
charge [10]
Figure: Machine Learning Server and its output[11]
3. CONCLUSION:
So, this IoT controlled quadcopter is more efficient in many
ways. All these machine learning algorithms can be used in
here by using this technique. As it transmits all its data to the
server and the machine learning server will to all processing
and just give a particular output to it .so the raspberry pi
doesn’t have to gain huge processing power it can just
execute certain command send from the server or theadmin.
This can be used in such a huge extend of object finding,
locating certain people. It can move freely and controlled
globally so it can be used in large ranges. It is more effective
than a normal CCTV camera to detect people and more
efficient in obstacle avoiding. All those ways of how this
model can be implemented are given in here. After all, it can
extend the era of machine learning technologytosucha great
extent.
REFERENCES
[1] IoT Based Agriculture Monitoring and Smart Irrigation
System Using Raspberry Pi. (2018). International
Research Journal ofEngineeringandTechnology(IRJET),
[online] 05(01), pp.1-2. Available at:
https://irjet.net/archives/V5/i1/IRJET-V5I1307.pdf
[Accessed 16 Feb. 2020].
[2] ESP8266 serial Wi-Fi wireless Transceiver
Module for IoT, ESPRUINO-Wireless.
[3] IoT Based Agriculture Monitoring and Smart Irrigation
System Using Raspberry Pi. (2018). International
Research Journal ofEngineeringandTechnology(IRJET),
[online] 05(01), pp.1-2. Available at:
https://irjet.net/archives/V5/i1/IRJET-V5I1307.pdf
[Accessed 16 Feb. 2020].
[4] G. Bradski, A. Kaehler, 2008, Learning OpenCVComputer
Vision with the OpenCV Library, O'Reilly
[5] S. Kakade, NJ Uke, Real Time human detection using
covariance matrices as human descriptor , Proceedings
of
4th International Conference on Computer and
Automation Engineering (ICCAE), 2012
[6] Cultivating Research in Computer Vision within
Graduates and Post-Graduates using Open Source.
(2012). International Journal of Applied Information
Systems (IJAIS), 04(01), pp.1-4.
[7] Big Data, Machine Learning and the BlockChain
Technology: An Overview. (2018). International Journal
of Computer Applications (0975 - 8887), 180(20), p.3.
[8] Sodhi, Pinky and Awasthi, Naman and Sharma, Vishal,
Introduction to Machine Learning and Its Basic
Application in Python (January 6, 2019). Proceedings of
10th International Conference on Digital Strategies for
Organizational Success. Available at
SSRN: https://ssrn.com/abstract=3323796 or http://dx.
doi.org/10.2139/ssrn.3323796
[9] THE MODEL OF DATA DELIVERY FROM THE WIRELESS
BODY AREA NETWORK TO THE CLOUD SERVER WITH
THE USE OF UNMANNED AERIAL VEHICLES.
(n.d.). Proceedings 30th European Conference on
Modelling and Simulation ©ECMS ThorstenClaus,Frank
Herrmann, Michael Manitz, Oliver Rose (Editors).
[online] Available at: http://www.iotlab.ru[Accessed25
Feb. 2020].
[10] microsoft. (n.d.). Microsoft Machine Learning Server 9.4
is now available. [online] Available at:
https://cloudblogs.microsoft.com/sqlserver/2019/07/3
0/microsoft-machine-learning-server-9-4-is-now-
available/ [Accessed 25 Feb. 2020].
[11] Configuring Microsoft Machine Learning Server to
operationalize analytics (One-Box Configuration)
(Linux). (2018). [image] Available at:
https://github.com/microsoft/microsoft-
r/tree/master/mlserver-arm-templates/one-box-
configuration/linux [Accessed 25 Feb. 2020].

IRJET - IoT based Facial Recognition Quadcopter using Machine Learning Algorithm

Recommended

More Related Content

What's hot (20)

Similar to IRJET - IoT based Facial Recognition Quadcopter using Machine Learning Algorithm (20)

More from IRJET Journal (20)

Recently uploaded (20)

IRJET - IoT based Facial Recognition Quadcopter using Machine Learning Algorithm