Virtual Trial Room Application

Dayasagar A S1 , Kushal S2 ,Sudeep Gowda C K3 , Sudharshan L4 ,Jagadeesh B N5

RNS Institute of Technology, Bengaluru

auctions, as individuals increasingly

Abstract: gravitate towards these platforms to fulfill
Abstract: Long lines and wait periods have their product needs. This paradigm shift in
turned into turnoffs for consumers in the transactional behavior not only signifies a
fast-paced world of today, which affects prevailing trend but also offers significant
retail stores' total sales. Furthermore, convenience to consumers.
keeping track of goods in expansive retail
showrooms is a difficult task that requires a However, a noteworthy challenge emerges
lot of labor and raises costs. This research in the context of online apparel purchases,
suggests using augmented reality (AR) and where customers encounter limitations as
artificial intelligence (AI) to power virtual they cannot physically try on the products
trial rooms as a way to overcome these before finalizing their transactions. The
problems. sensory experience of trying on clothes
Customers may digitally try on clothing in plays a crucial role in influencing a client's
virtual trial rooms, doing away with the decision-making process. The absence of
necessity for real fitting rooms. This creative this tangible experience in online clothing
method uses AI and AR to deliver a quicker purchases can impact the customer's
and more engaging alternative to confidence and satisfaction.
conventional techniques. Consumers can see
how apparel will seem on them without Problem Statement
having to go to the trouble of actually trying
it on. Create and install an application for a virtual
trial room so that customers may virtually
KEYWORDS: Face detection, Haar try on clothes before buying them. The
classifier, superimposing, blending, system need to offer a realistic and engaging
positioning, online shopping, online simulation that lets users utilize augmented
auctions, visualisation, dressing simulation,
reality or other immersive technologies to
and customer experience are some of the
applications of virtual trial rooms and smart see how various outfits might appear on
systems. them.To improve the virtual shopping
experience, take into account elements like
INTRODUCTION precise size, realistic clothing movement,
The swift advancement of technology has and an intuitive UI. A range of clothing
profoundly impacted our daily lives, styles should be supported, and the solution
ushering in a rapid transformation in various
should be adaptable to ensure that it works
aspects. Notably, there has been a
substantial surge in the adoption of digital with many platforms, including tablets and
practices, such as online shopping and smartphones.
SOLUTIONS FLOW CHART

 Permit users to conveniently

maintain their profiles within the
program by creating accounts for
them. Incorporating personal
dimensions into this capability would
guarantee that every user has a
customized and unique experience.

 Include a smart augmented reality or

other cutting-edge technology-based
virtual try-on capability. This gives
consumers a realistic representation
of how various clothing items will
look on them by enabling them to
digitally try on the things. Make sure The system's algorithm is given the tools
the virtual try-on is easy to use and and capabilities it needs to handle
provides a thorough view of the challenging jobs by utilizing these Python
chosen clothing items from different packages, guaranteeing efficiency and
perspectives. effectiveness in the creation and use of
machine learning and deep learning
algorithms.
BLOCK DIAGRAM
USED PROGRAMMING LANGUGE

The Python programming language was

used to create the system's whole algorithm.
Python is a high-level, interpreted
programming language that is used
extensively to create sophisticated deep
learning and machine learning algorithms by
employing different packages.
 tracking the movement and gestures of
hands in real-time. MediaPipe Hands
OPEN CV employs a machine learning pipeline that
integrates multiple models, working
collaboratively to enhance the accuracy and
efficiency of hand tracking.

RELATED WORK

The open-source computer vision library

OpenCV is well regarded and is mostly used
for image processing. When it comes to real-
time computer vision applications, it is well
known for its adaptability. C++ is used to
implement the library's main design.
Launched as an open-source project in 2011,
OpenCV provides a wide range of tools and Furthermore, the program continuously
features for various computer vision and computes distances in order to monitor
image processing applications. bodily mobility. An essential component of
this procedure entails:
MEDIA PIPE
Unit Conversion: The distances that are
initially measured in meters are those that
are computed using joint coordinates. A
conversion to pixel data is necessary in order
to map the data onto the user's body and
A Python library called MediaPipe was merge it with pixel-based data smoothly. A
created as a machine learning package and constant and exact transformation is ensured
comes with pre-installed TensorFlow-trained by using a conversion factor of 1 meter,
datasets. This library serves as a robust which is equivalent to 3779.527559055
foundation for applications such as sign pixels.
language understanding, hand gesture Joint Position Adjustment: At first, the
control, and augmented reality content application depended on the sensor's
overlay. MediaPipe's primary strength lies in identification of important places.
providing an optimal solution for hand and Nevertheless, because of their imprecision,
finger tracking. several joint locations presented difficulties,
One of the key functionalities of MediaPipe especially those related to the hip center, hip
is its utilization of deep learning techniques left, and hip right.
to infer 21 3D landmarks on a hand. This Several efforts were undertaken to address
capability is instrumental in accurately this problem, but adequate outcomes were
never achieved. The decision to establish Frame Specification: Frames per trigger and
new joint roles was taken after much frame capture intervals are defined at the
deliberation. These other points were added programming stage. The webcam takes a
to solve the issues of imprecision and picture as soon as it finds the desired frame.
alignment. They were placed strategically to In order to provide a regulated and
the left, right, or below from the original synchronized picture acquisition for later
positions. processing, this step is necessary.

DATA FLOW DIAGRAM TABLE 1: The Basic Commands Used

COMMANDS FUNTIONS

Imread To read an image

Imshow To display an image

Imaqreset To reset the camera

rgb2gray To convert a colour

image to gray scale
image
IMAGE CAPTURING
gray2rgb To convert a gray scale
image to colour image
The project's primary method is picture
acquisition, which uses an external or
integrated webcam to record hand motions.
TESTING AND TRAINING
This is accomplished by using MATLAB's
Image Processing Toolbox, which makes Correlation operations are used to compare
later phases of the system easier. matrices that are created from gestures that
are captured and formatted into a matrix
Camera Information Retrieval: It is essential
with those that are kept in the database. The
to obtain information about the attached
two main components of the image
camera before beginning programming. For
processing pipeline are testing and training.
this operation, the 'imaqhwinfo' (Image
Acquisition Hardware Information) Training - Database Creation: During this
command is used. Furthermore, in order to stage, the system concentrates on building a
precisely identify the camera, the device's large database.
ID may be acquired by using the MATLAB
command window command
'imaqhwinfo('winvideo')'.
The following are the steps that make up this
phase:
CONCLUSION:
Pre-processing: Dynamic changes to
The cloth model has been integrated with the
brightness, contrast, and sharpness are made
enhanced performance joint positions,
to the acquired gesture picture during the
transforming this application into a highly
pre-processing phase. This improves the
functional virtual fitting room that provides
input photos' general quality and uniformity
users with a smooth and engaging
and gets them ready for further analysis.
experience. Among this application's
Feature Extraction: After pre-processing, the primary features are:
system uses the gesture photos to identify
Human Measurement Generation: By
distinguishable characteristics. The process
photographing the user's body posture in
of feature extraction entails identifying
front of the Kinect sensor, the program
pertinent patterns or traits that add to the
creates precise human measurements. This
distinctiveness of every motion. These
fundamental stage guarantees accurate
characteristics are essential for
fitting and a customized virtual experience
differentiating motions throughout the
based on the individual body measurements
comparing procedure.9.
of the user.
Database Population: Every gesture that has
Realistic and Flexible Cloth Model: By
been preprocessed and enhanced with
allowing users to "wear" realistic and
features is entered into the database. This
flexible cloth models digitally, the virtual
extensive database acts as a guide for further
fitting experience becomes more realistic.
experimentation and contrast.
The fabric model dynamically adjusts to the
user's body to give a realistic portrayal of the
selected outfit as it would seem in real life.

User-Friendly Body Motion-Based GUI:

The program has a body motion-based
graphical user interface (GUI) that is both
simple to use and intuitive. The virtual
fitting room is simple for users to manage
and navigate around using their natural body
motions. This creative method improves the
user experience overall and provides a
convenience layer.

REFRENCES:

[1] Higgins, K. R., Farraro, E. J., Tapley, J.,

Manickavelu, K., & Mukherjee, S. (2018).
U.S. Patent No. 9,898,742. Washington, DC:
U.S. Patent and Trademark Office.

[2] Isıkdogan, F., & Kara, G. (2012). A real

time virtual dressing room application using
kinect. CMPE537 Computer Vision Course
Project.

[3] Korszun, H. A. (1997). U.S. Patent No.

5,680,528. Washington, DC: U.S. Patent and
Trademark Office.

[4] Protopsaltou, D., Luible, C., Arevalo,

M., & Magnenat-Thalmann, N. (2002). A
body and garment creation method for an
Internet based virtual fitting room. In
Advances in Modelling, Animation and
Rendering (pp. 105-122). Springer, London.

[5] Steermann, M. C. (2014). U.S. Patent

Application No. 14/215,649.