Chapter client in route without the assistance of

Published by admin on

Chapter 1



In this chapter, a
detail introduction of the fundamental concepts related the presented project
is discussed. In the latter half of the chapter, problem description and
project objectives are described.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

1.1  Background:

In our planet of 7.4
billion people, 285 million are outwardly hindered out of whom 39 million
individuals are totally visually impaired i.e. have no vision at all and 246
billion individuals have mellow or serious visual impedance. It has been
anticipated that by 2020 these numbers will raise to 75 million visually
impaired and 200 outwardly weakened individuals. Since the 1970, question
acknowledgment advances have developed to a time when energizing applications
are getting to be plainly workable for visual substitution. Indeed, industry
has made an assortment of PC vision items and administrations by growing new
electronic guides for the visually impaired with a specific end goal to defeat
the challenges that the pooch and stick don’t react. Building up an instrument
for the outwardly debilitated individuals isn’t an as of late risen issue. In
any case, building up a PC helped device is an as yet creating region. The
point of every one of these frameworks is to help the client in route without
the assistance of a moment individual. There are a few works utilizing PC
vision systems. Yet, there is no current strategy that assistance to illuminate
the every single fundamental need of visually impaired individual.


Identifying separation
sensors, for example, RFIDs, ultrasonic sensors and infrared sensors, are
typically used to alter direct sticks Regarding the RFID detecting strategy, a
RFID peruser module is introduced on a guide stick, and an expansive amount of
RFID labels are situated underground to position and route. Such assistive
gadgets are not broadly utilized and are hard to apply, in actuality,
circumstances; what’s more, without expansive amounts of RFID chips being
situated underground ahead of time, the framework can’t work. Concerning and
infrared sensors, they are shabby and effortlessly executed into a guide stick.
These sorts of sensors can correctly decide the separation and inform the
outwardly weakened individuals there are deterrents in front, yet they can’t
perceive the impediment classifications. In further developed plan, numerous
infrared sensors can distinguish extraordinary cases, for example, stairs, yet
at the same time can’t identify the more unpredictable state of the impediment.


There were bundle of
endeavors made in mid 80’s and 90’s every one of them attempted to get the
comparable approach. Be that as it may, utilizing that approach did not appear
to demonstrate gainful in light of the fact that distinctive of pictures of
similar articles did not appear to be identical, so the outcome is constantly
frightful. This was the errand the people can just do. Be that as it may, in
1998 LeCun presented a convolutional neural system equipped for ordering system
with 99% exactness which demonstrated CNN learned highlights independent from
anyone else. Profound convolutional systems have a long history in PC vision,
with early cases indicating victories on utilizing regulated back-spread
systems to perform digit acknowledgment (LeCun et al., 1989).

Software Requirement specification

Software Requirement Specification (SRS) is a description of a software system
to be developed. It requires the basis for an agreement between customers and
contractors. It enlists enough and necessary requirements that are required for
the project development. It lays out functional and non-functional
requirements.   Functional

. Camera

                 It should be of high definition.


is a wearable technology so; frame will be used which should be of best

Headphone / Audio generator:

should be of moderate frequency that is safe for both adults and children.


software has been selected due to the project analysis characteristics.
Multiple image analysis functions have been built into this software.


is used for numerical competition
Non-Functional Requirements

The software algorithms should be of high
efficiency to aid in quick response time.

The height and width at which an object
will be considered an obstacle will need to be determined empirically.

It is ensured that there are good
lightening conditions.

It is easy to use.

It is affordable.


Semantic Technology

Semantic innovation is an arrangement of
techniques and instruments that give propelled intends to ordering and handling
information, and additionally to discover connections inside fluctuated
informational collections.

In our project, Deep Convolutional Neural
Networking is used. It is implemented in MATLAB for segmentation of objects.
Different packages of Python such as

 numpy /scipy
have adapted for efficient numerical competition and training the network.


1.2 Problem

Outwardly weakened individuals for the most part
have issues strolling and maintaining a strategic distance from hindrances in
their day by day lives. Customarily, such individuals utilize direct sticks to
identify deterrents before them. Along these lines, outwardly impeded
individuals can’t precisely recognize what kinds of impediments are before them
and should just rely upon control sticks. Assistive gadgets for route for
outwardly weakened individuals still concentrate on area and separation
detecting, however can’t educate clients concerning the kinds of obstructions
before them. Separation detecting can’t give extra data to help outwardly
weakened individuals to comprehend their environment. Along these lines, the
practicability of such assistive gadgets is low. Late advances in minimal
effort wearable PCs opens up new conceivable outcomes for the improvement of
inventive visual guides. The primary motivation to construct such wearables is
to diminish hindrances in life of outwardly hindered individuals. Perceiving
diverse items and content is most imperative and addressed part to implement
savvy glasses. Our wearable gadget proposes a route framework for outwardly
debilitated individuals.

1.3 Objectives

 Our fundamental target is


•           to
build up an extraordinary picture acknowledgment calculation for shapes and
hues for constant application utilizing MATLAB.


•           to
examine the execution of picture acknowledgment calculation in term of
exactness and time handling.


•           to
build up a calculation to change over perceived picture to voice utilizing


•           to
examine the execution of picture to voice transformation calculation.


•           to
test the execution of the shut circle interface for the picture and sound
preparing converter framework.


•           to
create graphical UI (GUI) of the picture to voice converter for instance of
client finding.


1.4 Project

This device is gaining a place in
our society and reshaping our technological scene. Our initial step is to make
a device that will recognize different objects like currency notes and fruits
stored in database. In long term perspective, virtual eyes can be made in

Smart glasses for blind will have camera
that will scan nearby objects.

all indoor objects will be identified.

The range of objects are currency notes,
table, bed, chair glass, plate, jug, door, windows, food items (banana, apple,
orange, bread, biscuits).

To acquire different question
acknowledgment deliberately, a profound learning strategy is actualized to
naturally learn and separate component esteems, radically shortening the
extraction time of protest highlights. For picture acknowledgment,
Convolutional Neural Networks (CNNs) are the best among all present profound
learning structures. CNNs have the upsides of diminishing picture clamor,
expanding signal highlights, and streamlining neural systems. Accordingly, CNNs
are being connected broadly in profound learning, and they speak to the most
well known model design connected by a great many people.Vision and object
learning will be based on deep learning algorithm. A deep convolutional network
is used for it which is first trained in a fully supervised setting using state
of art method. We will then extract various features from this network, and
evaluate the desired result for these features on generic vision.

First vision will be performed through
android mobile app and using mobile phone camera. When the system begins to operate required image is sent to
the backend server to be processed. The backend server uses the faster region
convolutional neural network algorithm to recognize multiple obstacles in every

Finally, smart glasses will be equipped
with camera and AI software to see objects.



1.5 Gantt Chart

Gantt chart is
a type of bar chart that illustrates a project schedule. Gantt charts
illustrate the start and finish dates of the terminal elements and summary elements
of a project. Terminal elements and summary elements comprise the work break
down structure of the project. 





1.6 Tools
and technology used

main tools used for developing smart glasses that will assist blind people

Mat lab 9.3

Python 3.6.4

R language

Deep convolutional neural network

Alex Nets



Categories: Industry


I'm Iren!

Would you like to get a custom essay? How about receiving a customized one?

Check it out