DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING

FACULTY OF ENGINEERING & TECHNOLOGY

A PRESENTATION ON

MULTIPLE POSE VIRTUAL TRY ON BASED ON 3D CLOTHING RECONSTRUCTION

PRESENTED BY : MD SAMEER SHAIKH USN : 19KB02BS[030]

MOHAMMED MUSTAFA MOHIUDDIN USN : 19KB02BS[040]
SYED UMER FAIZAAN USN : 19KB02BS[079]
SYED ZUBAIR PASHA USN : 19KB02BS[080]

UNDER THE GUIDANCE OF : PROF. AMREEN ANJUM

 CONTENTS
 INTRODUCTION
 PROJECT DEFINITION
 OBJECTIVE
 LIMITATIONS
 LITERATURE SURVEY
 METHODOLOGY
 EXISTING SYSTEM
 PROPOSED SYSTEM
 SYSTEM DESIGN
 SYSTEM REQUIREMENT
 IMPLEMENTATION
 CONCLUSION & FUTURE SCOPE
 REFERENCES
 INTRODUCTION

 Online shopping and e-commerce are rapidly increasing worldwide due to

their efficiency and convenience. However, in the case of fashion items,
especially clothing, a fair percentage of purchased products are returned
due to a mismatch in style and size.

 Virtual Try-On (VTON) technology was invented to narrow the gap between
these online viewing and purchasing experiences by creating virtual images
and real time videos that help users have a more thorough and realistic
shopping experience

 Multiple pose VTON (MPVTON) systems can generate virtual fitted images
in multiple poses with different try-on clothes from a single customer image
as input.
DEFINITION OF
PROJECT:

● Buying apparel through e-

commerce has certain limitation
of trying apparel before buying it.
Provide a virtual room to try
apparel through e-commerce
websites before buying it.
OBJECTIVE

&
MO DER NE
# SIMPL E
 The main objective of this
project is to satisfy
customers while shopping
online to find their perfect
combination of clothes.

 To promote organizations
using Digital Marketing.
LIMITATIONS

 It is not cent percent accurate.

 Requires fast internet

connectivity.

 Requires good lighting.

 LITERATURE SURVEY

 RESEARCH PAPER-1
Author: Huabai liu.
Title: Computer 5G Virtual Reality Environment 3D Clothing Design.
Year: 2022

OVERVIEW: This paper analyzes the 3D clothing design process in detail through .
demand analysis

PROPOSED MODEL: Author Proposes a new three-dimensional cutting method.

Based on the constant distance from the center key point to the boundary key
point before and after the 3D garment is expanded, a geometric expansion
algorithm based on the calculation of the length of the 3D B-spline curve is
proposed.

LIMITATION: Use of 5g frequency can damage human skin and brain cells and is
also not appropriate for animals. It may reduce the life span and can cause
various diseases like skin cancer.
RESEARCH PAPER-2
Author: Liu Fengyi and Siru Liu
Title: 3D Garment Design Model Based on Convolution Neural Network and
Virtual Reality.
Year: 2022

OVERVIEW: The development of virtual reality technology has promoted the

unceasing reform and development in the field of fashion design.

PROPOSED MODEL: In this paper, a three-dimensional clothing design model

based on convolution neural network and virtual reality is established
through the recognition of human body morphological features and the
analysis of clothing sample data. This model provides a new method and
approach for clothing product design based on virtual reality.

LIMITATION: Intelligent design bureaus of clothing are limited to the

exploration of shallow layer neural network, intelligent design of machine
learning is less, so, under the background of artificial intelligence to clothing
of intelligent design method is a new breakthrough.
RESEARCH PAPER-3
Author : Sara Rankohi & Lloyd Waugh
Title: Review and analysis of augmented reality literature for construction
industry.
Year: 2013
OVERVIEW: Research has identified various beneficial capabilities for
augmented reality technologies in the AEC industry such as virtual site
visits, comparing as-built and as-planned status of projects, pre-empting
schedule disputes, enhancing collaboration opportunities, and planning/
training for similar projects.

PROPOSED MODEL: The complex nature of the architecture, engineering,

construction, and facility management (AEC/FM) industry and its high
demand for access to information for evaluation, communication and
collaboration, increases the industry’s need for information technologies.

LIMITATION: Bulky and expensive headsets with a limited field of view (FoV)
, security concerns when AR data is manipulated toinfluence worker
decisions, a high and expensive learning curve to use, and a lack of truly
precise spatial location systems for AR objects.
RESEARCH PAPER-4
Author: Rshami S. Shinkar & Nagaraju Bogiri
Title: A Real Time Virtual Dressing Room Application using Opencv.
Year: 2019

OVERVIEW: The real time clothes trying having long waiting queues for trial rooms is a
problem faced at many places and store. In this system, by using real time system cloth
processing is done by data drive approach which is worn by human. To give exact look to
skeleton, by using height, skin color, it starts creation of clone person which is prior to real-
time simulation.

PROPOSED MODEL: To make platform independent approach and portable approach we

should implement this by using OpenCV. Our approach is as follows: By using depth and
user label data, by video stream user should be extracted. According to the skeletal tracker,
Positioning of the 2D cloth models. From the sensor distance should be scaled between
body joints and user.

LIMITATION: There is no attainable for use a similar variation to the pictures instead of
straightforward scale-rotate ability for matching many joints altogether though it'd
necessary additional estimation. Many pictures at different angels cannot be used because
of it, it is not feasible for creating more practical video stream.
 METHODOLOGY

For building an end-to-end projection using Augmented reality

application it is necessary to go through the following phases listed
below:
1. Model construction .
2. Selecting AR Powered Engine.
3. Model Licensing.
4. Application Interface .
 EXISTING SYSTEM
The existing system of model projection using augmented

&
MO DER NE
reality has a very tidy process of hardware resources to be of

# SIMPL E
very good efficiency as We need cameras of better resolution
and currently used Augmented Reality applications are very
heavy in terms of size and space, A general smart phone with
minimal specification can hang and lag with the application
leading to improper projection or No projection at all.

Some of the drawbacks are:-

 Needs a flat Surface to project the model.
 Better hardware resources are required.
 Need proper stable camera setup.
 Cannot implement our own model or build by our own.
 Need a higher resolution camera.
 Not applicable for all smart phones.
PROPOSED SYSTEM
The proposed system is a Real Time Model Projection
using Augmented Reality where we are trying to build
an mobile application which is resource friendly to
make projections of some pre install models and
project them on real surfaces such as Floors, walls,
rough surfaces Etc. Aim of this project is to make a
light weight application suitable for all types of smart
phones and ability to make their own model
projection of their choice.

THE APPLICATIONS INCLUDE:

 Executing the application with low hardware
smart phones.
 Available for across all android phones.
 Takes lesser space in memory.
 Ability to introduce custom models for projection.
 Could work with uneven and rough surfaces as
well.
SYSTEM DESIGN
SYSTEM REQUIREMENTS
Software Requirement:
 Cmake (3.12.0)
 Unity
 Opencv (3.4.2.17)
 Scipy (1.0.0)
 Cascade trainer gui (1.8.0)
 Tkinter canvas (8.6.8)
 Numpy (1.18.1)
 Anaconda (4.8.2)
 Python (3.7.4)
 Flask Web framework (1.1.1)
Hardware Requirements
 For this project we need high configurations
system because such tasks in this project
requires more CPU computation and high
graphic support to train haar cascade dataset
and for more accuracy we need to increase
number of EPOCH in training of convolutional
neural network.
 Once the dataset and network models are
trained then we can run python code on normal
machine also.
 IMPLEMENTATION
User Extraction/Detection from the Input image : By separating user
area user allows creating environment and background from the image
and layered it onto a virtual manner in the user interface.

Human Body Detection: The main function of this step is to determine

whether human body appears in a given image, and where these are
located at.

Kinect Skeletal tracking: Large number of dimensions is described in the

skeletal tracking. Skeletal tracking dimensions describe unique
individuals, shapes, size, hair and clothing etc.

Face detection using AR from live camera: It uses Haar-Like features

algorithm, this algorithm created by Voila and Jones is used for finding
the location of human faces in a frame, the human faces has some
universal properties like the nose region is brighter than other regions
like the eye region and the eyes region is darker than the pixels around it.
RESULTS
FUTURE SCOPE
 After detecting the physique of the
targeted person further it can be
improved by application of three tools
which will help to set the clothes on target

 PAN (pose alignment network)

 TRN (texture refinement network)
 FTN (fitting network)
 CONCLUSION
Recent advancements in Augmented Reality are helping ease
the fashion industry's transition from customary stores into
modern online shops equipped with high-tech features such as
virtual try-on and fashion synthesis systems.

we demonstrate a completed process for garment

computational representation, computational styling and user
preference training and modeling. Our model and
implementation show great performance on different datasets
and proved capable of modeling users’ preferences.
 REFERENCES
 H. Liu, “Computer 5G virtual reality environment 3D clothing
design,” Mobile Information Systems, vol. 20, no. 13, pp. 5–11,
2021.
 Y. Hong and Y. Ge, “Design and analysis of clothing into
account unity's immersive virtual reality in an artificial
intelligence environment,” Computational Intelligence and
Neuroscience, vol. 52, no. 11, pp. 52–61, 2019
 Y. L. Jia, “The research of dance costume design based on
virtual reality technology,” Advanced Materials Research, vol.
18, no. 21, pp. 15–34, 2018.
 https://www.google.com/ml
 https://en.wikipedia.org/wiki/Augmented_reality
 https://store.unity.com/download?ref=personal
THANK
YOU