VIRTUAL TRY-ON
(FASHION RECOMMENDATION
EXPERT SYSTEM)
DS - AI
Vibhuti Sahukari
Janvi Tiwari
BFT-4
WHAT IS AN EXPERT SYSTEM?
Medium of sending the information back to the user- I think neon is in Suggesting new
User Interface trend nowadays! information- Inference
Engine
Knowledge from an User getting the information
Expert (here, fashion asked in a particular format
consultant) including, advice as text,
rating, or image solution
etc.
Data stored in the back end-
Knowledge Base
Giving the required information to the user in the certain medium
as asked (in this case, an advice)
Just by Definition
AN EXPERT SYSTEM IS?
An expert system is a computer program that uses artificial intelligence (AI)
An expertto
technologies system is athe
simulate computer
judgmentprogram that uses
and behavior of artificial
a human or intelligence (AI)
an organization
technologies that
to simulate the judgment
has expertise and behavior
and experience of a humanfield.
in a particular or an organization
that has expertise and experience in a particular field.
It uses the principles of machine learning and Artificial Intelligence to provide
Itsolutions
uses the to
principles of machine
the problems asked learning andexisting
from a pre- Artificial
setIntelligence
of knowledge to provide
base
solutions
provided to the
by the problems
experts in theasked from field
particular a pre-
in existing setthe
a way that of knowledge base
system is capable
provided by the experts inof the particular
learning withfield in a way that the system is capable
experience.
of learning with experience.
Knowledge Base
Expert Knowledge Inference Engine User
User Interface
Source: https://www.techtarget.com/searchenterpriseai/definition/expert-system
FASHION
CONSULTANT
OR
STYLIST
SYSTEMS
Internet Based Human Stylist Consultant
Systems put the communication between clients
and stylists on the Internet, by giving various
styling tips and recommendations according to
the season, trend, and also body type of people,
in the form of data fedby experts into the expert
system softwares.
They improve the flexibility and accessibility of
styling work
FLOW CHARTS
Source: Research Paper- The Design and Implementation of an Intelligent ApparelRecommend Expert System by Hindawi
FLOW CHARTS
Explaining the system
Source: Research Paper- The Design and Implementation of an Intelligent ApparelRecommend Expert System by Hindawi
Gesture Recognition NLP 3D Body Scanning Virtual Clothing-CAD
Augmented Reality
a fashion recommendation ES
COMPONENTS
Machine & Deep Camera- Face and Age
Speech Recognition Neural Networks and
Learning Recognition
Computer Vision Predictive Analytics
Smart Magic Mirrors - Virtual Try
On - Offline
Mobile Virtual Try On - Online
APPLICATIONS
of a Fashion Recommendation ES
Hands free Image Capturing and
Recommendation devices
Make up and hair Virtual Try on -
Online
CONFIGURATION MAP OF AR BASED MAGIC MIRROR
The display first scans the image of the
person standing in front of it by the camera
present, then it creates either a "VIRTUAL
AVATAR" of the person or simply using
"COMPUTER VISION" gains knowledge
about the size and measurements of the
person.
It then displays garments according to the
ones asked by the user in the form of
"VIRTUAL CLOTHING" which he/she can
navigate through using "GESTURE
RECOGNITION".
The user can then turn on the Style check
features for style recommendations and
outfit check options, while the system uses
"COMPUTER VISION" again to check for the
"ATTRIBUTES" including colour, season,
texture and pattern etc. for the garment.
Source: Research Paper- Augmented Reality Fashion Apparel Simulation using a Magic Mirror
RECOMMENDATION
RECOMMENDATION
FASHION
FASHION
Source: Research Paper- Artificial Intelligence and fashion industry by Luiss
The five major components include
Initialization program
Database
Style Engine
Learning Components
User Interface (UI).
The main system is a Java Servlet application written in
Java and HTML, which uses MySQL as the back-end and
a fashion recommendation ES runs on a web server.
STYLE ME It first takes an initial quiz with 8 questions to know the
person.
home based virtual try on It already has a database containing the styling rules
along with the tables for all inlcuding clothing, shoes and
pairs.
The user is displayed with a pair from the clothing and
shoes tables using ANN and is stored in the pairs table.
They can easily reset the system for another pair option
which will also be added to the pairs table and with the
deep learning feature the system learns from the
feedbacks and choices of the users.
System's Working
FLOW
CHART
Source: Research Paper- AN ARTIFICIAL INTELLIGENCE BASED CLOTHING FASHION STYLIST by Haosha Wang
AMAZON
ECHO LOOK
a personal fashion stylist
Launched on June 6, 2018 on
Amazon store
Source: Research Paper- Artificial Intelligence and fashion industry by Luiss
Introducing
STYLE CHECK
an Amazon Echo Look's feature
This system uses Machine Learning Algorithms
and advice from fashion specialists.
Submit 2 photos for a second opinion on
which outfit looks better based on fit, color,
styling and current trend, etc.
The decisions get smarter through our
feedback and input from the team of
experienced fashion specialists.
Source: Research Paper- Artificial Intelligence and fashion industry by Luiss
The app will suggest items that pair
with pieces you already own, organize
your photos into collections, search by
color, and receive more specific
feedback.
One style check result was - "outfit of a
black cardigan, pattern blouse and
burgundy pants to those same pants
with a different sweater because the
patterns and shape of the outfit work
better"
VIRTUAL TRY ON
Source: Research Paper- Artificial Intelligence and fashion industry by Luiss
LENSKART'S VIRTUAL TRY ON
Source: Lenskart Website
VIRTUAL HAIR
TRY ON
Source: Loreal Pairs Website
The phrase ‘predictive analytics’ includes a group of techniques from
statistics to
machine learning that uses historical data to make predictions using
models. PREDICTIVE FIT
A model in predictive analytics is an algorithm that uses the past to
predict the future. ANALYTICS
Deep-learning algorithms, based on neural networks, offer
&
considerable help in demand forecasting. TREND FORECASTING
Long-short -term memory (LSTM) models are used for time-series
forecasting or a transfer learning model helpful for forecasting using
small data sets. AI could also be particularly helpful in tracking fashion
trends.
Source: Research Paper- Smart Fashion: A Review of AI Applications in the Fashion & Apparel Industry
HOW IS IT DIFFERENT FROM AI
A Fashion Recommendation Expert System is different from AI in such a way that that
AI only provides garment recommendations according to the previous activity of the buyers
but
Expert System will recognize the body structure, age and face of the person, through computer vision
will analyse the measurements to display the garments according to their body type along with
ananlyzing the fabric type, print, colour and drape according to the person's liking and the season,
occasion or time of the day specified.
The system is capable of providing the user with different pairing of outfits and accessories according to
the rules and data fed in by fashion styling experts along with the forecasting and feedbacks done and
collected resp by the system, through various principles of machine learning, ANN, and pridictive fit
analytics.
ADVANTAGES DISADVANTAGES
of an E-Fashion Recommender
Virtual Offline Try on feature helps the user to Styling Tips provided by the system can
get a garment without having to actually try it sometimes be outdated with the current
on physically. trend if the rules and algorithm is not
Expert styling tips can be easily received by revised regularly.
any common person without having to pay for The outfit recommended by the system
consultants. may/may not be according to one's style
The face, age and size recognition feature or liking.
allows the user to get a more appropriate size It might get difficuilt for the system to pair
according to their body type. clothes, accessories, hair, and makeup
Trying out the clothes by mixing and matching for every individual according to their taste
is made easier with the virtual assistance. of dressing along with keeping up with the
One can see how a virtual garment is looking trends.
on them and then can physically have a feel Stores with magic mirrors have to keep the
test of it from the store they are making a systems updated towards inventory
purchase in. management of each garment part.
a note to
THE FUTURE
As a future task, a convergence-type fashion coordination contents system will be
developed by adding real-time customized fitting functions that apply VCA (Video
Content Analysis) technology.
Also, our data structure and styling rules are too simple for practical use. We require a more holistic
data set containing the information about the fashion trends from various sources altogether so that a
collective environment is created.
The change towards smart and virtual technology in the fashion and garment industry is still in progress
and a lot is yet to be achieved with the "Smart and Virtual Garment" that the users can fill their closets
with.
The industry is still learning to accommodate the applications of AI in this field and will slowly move
towards a more digital environment with the future generations and developments.
REFERENCES
1-Source: https://www.techtarget.com/searchenterpriseai/definition/expert-system
2-Source: Research Paper- Smart Fashion: A Review of AI Applications in the Fashion &
Apparel Industry
3-Source: Loreal Pairs Website
4-Source: Research Paper- Artificial Intelligence and fashion industry by Luiss
5-Source: Research Paper- Artificial Intelligence and fashion industry by Luiss
6-Source: Research Paper- Augmented Reality Fashion Apparel Simulation using a Magic
Mirror
7-Source: Research Paper- The Design and Implementation of an Intelligent
ApparelRecommend Expert System by Hindawi
THANK YOU
