PES UNIVERSITY

(Established under Karnataka Act No. 16 of 2013)

100-ft Ring Road, Bengaluru – 560 085, Karnataka, India

Capstone Project Report Phase-2

on

Find your Fit

Submitted by

Aishwarya H R - (PES1PG22CA011)
Feb. 2024 – June 2024
under the guidance of

Guide Details
Dr. S Thenmozhi

Professor
Department of Computer Applications,

PESU, Bengaluru – 560085

 FACULTY OF ENGINEERING
DEPARTMENT OF COMPUTER APPLICATIONS
PROGRAM – MASTER OF COMPUTER APPLICATIONS

CERTIFICATE
This is to certify that the project entitled

Find your Fit

is a bonafide work carried out by

Aishwarya HR - (PES1PG22CA011)
in partial fulfillment for the completion of Capstone Project Phase-2 work in the
Program of Study MCA under rules and regulations of PES University, Bengaluru
during the period Feb. 2024 – June 2024. The project report has been approved as it
satisfies the academic requirements of 4th semester MCA.

Internal Guide Chairperson Dean- Faculty of

Engineering & Technology
Dr. S Thenmozhi Dr. Veena S Dr. B K Keshavan
Professor Professor and Chairperson Professor and Dean
Department of Computer Applications Department of Computer Applications PES University
PES University PES University Bengaluru - 560085
Bengaluru - 560085 Bengaluru - 560085
 DECLARATION

I, Aishwarya H R, bearing PES1PG22CA011 hereby declare that the Capstone

project phase-2 entitled, Find your Fit, is an original work done by me under the
guidance of Dr. S Thenmozhi, Professor, PES University, and is being submitted in
partial fulfilment of the requirements for completion of 4th Semester course in the
Program of Study MCA. All corrections/suggestions indicated for internal assessment
have been incorporated in the report. The plagiarism check has been done for the
report and is below the given threshold. I further declare that the work reported in this
project has not been submitted and will not be submitted, either in part or in full, for
the award of any other course.

PLACE: Bangalore

DATE: 19-06-2024

Aishwarya H R
 ACKNOWLEDGEMENT
The successful completion of any task would be incomplete without mentioning the people
who made it possible.

I would like to express my deep sense of gratitude to the honourable Vice Chancellor of
PES University, Dr. J Suryaprasad, for giving the opportunity to work on this project.

I take this occasion to thank my sincere and heartfelt thanks to Chairperson, Department
of Computer Applications Dr.Veena S, for her motivation, support and for providing a
suitable working environment.

With a great pleasure, I express my sincere gratitude to my guide Dr. S. Thenmozhi,

Professor, Department of Computer Applications, for providing me with right guidance
and advice at the crucial junctures which helped me in completing the project work on time.
I am wholeheartedly thankful to her for giving me valuable time, suggestions and for
showing me the right way in completing my project successfully.

I extend my sincere thanks to our project coordinator Mr. Tamal Dey, Assistant
Professor, Department of Computer Applications, for providing schedule and timelines
and documenting information about project.

I would like to express my gratitude towards my parents for their kind cooperation and
encouragement which helped me in the completion of this project.

My thanks and appreciation to my friends for the constant support and people who have
willingly helped me out in this project.
 ABSTRACT

Fashion is a means of expressing one’s self, emotions, and attitude. Online sales are a
rapidly growing element for many businesses. However, few industries utilise e-
commerce quite as much as the fashion industry. Clothing and footwear being high
involvement products in fashion industry, it is very difficult to make sales online as size
and fit are the most important criteria which influences the customer satisfaction. The
need for this type of system is that the present world has gradually begun to digitize
and everyone prefers online shopping as it saves lot of time. Especially in footwear
selection, many people go wrong because the foot size of a person varies from brand to
brand. Hence this type of detection will help people to find their perfect fit that allows
a smooth transaction from both ends.

This project proposes a Footwear Size Detection and Recommendation system which
uses OpenCV (Open-Source Computer Vision Library) based logic for calculation of
foot size. The web application is highly beneficial for customers who shop online
regularly. The user can upload their foot image and the logic will calculate the foot size
and display the size in Euro, UK, US and centimetres. This will serve as an advantage
for user to know their foot size in different scales. Recommendation system is also
developed in which the user can select footwear from different brands according to their
preference. Based on the user selection, the page will be directed to that brand and type
of footwear. The end users for this system will be customers who shop for footwear
online because it enables them to choose their perfect match without any hassle.
 CONTENTS
ABSTRACT
LIST OF FIGURES
LIST OF TABLES
1. INTRODUCTION .................................................................................................................. 1
1.1 PROJECT DESCRIPTION............................................................................................... 2
2. LITERATURE SURVEY .......................................................................................................... 5
2.1 DOMAIN SURVEY ....................................................................................................... 5
2.2 RELATED WORKS ........................................................................................................ 7
2.3 EXISTING SYSTEM ....................................................................................................... 8
2.4 TECHNOLOGY SURVEY ............................................................................................... 8
2.5 FEASIBILITY STUDY ..................................................................................................... 8
3. HARDWARE AND SOFTWARE REQUIREMENTS ................................................................ 11
3.1 HARDWARE REQUIREMENTS ................................................................................... 11
3.2 SOFTWARE REQUIREMENTS .................................................................................... 11
4. SOFTWARE REQUIREMENTS SPECIFICATION ................................................................... 13
4.1 USERS ....................................................................................................................... 13
4.2 FUNCTIONAL REQUIREMENTS ................................................................................. 13
4.3 NON-FUNCTIONAL REQUIREMENTS ........................................................................ 14
5. SYSTEM DESIGN ............................................................................................................... 16
5.1 ARCHITECTURE DIAGRAM........................................................................................ 16
5.2 BLOCK DIAGRAM / CONTEXT FLOW DIAGRAM ....................................................... 17
6. DETAILED DESIGN ............................................................................................................ 19
6.1 PROCESS FLOW DIAGRAM WITH DESIGN METHODOLOGY ..................................... 19
6.2 METHODOLOGY ....................................................................................................... 20
7. IMPLEMENTATION ........................................................................................................... 25
7.1 PSEUDOCODE ........................................................................................................... 25
7.2 SCREENSHOT ............................................................................................................ 31
8. SOFTWARE TESTING......................................................................................................... 38
9. CONCLUSION.................................................................................................................... 42
10. FUTURE ENHANCEMENT .............................................................................................. 44
11. APPENDIX A .................................................................................................................. 46
12. PLAGIARIASM REPORT ................................................................................................. 49
13. POSTER ......................................................................................................................... 52
 LIST OF FIGURES

5.1: Architecture Diagram ....................................................................................................... 16

5.2: Context Diagram ............................................................................................................... 17
6.1: Process Flow Diagram ...................................................................................................... 19
6.2.1: Input Image .................................................................................................................... 20
6.2.2: Pre-Processing ............................................................................................................... 21
6.2.3: K-Means Clustering ....................................................................................................... 21
6.2.4: Edge Detection............................................................................................................... 21
6.2.5: Contours Image .............................................................................................................. 22
6.2.6: Cropped Image............................................................................................................... 22
6.2.7: Overlay Image ................................................................................................................ 22
7.1.1 a: Home Page .................................................................................................................. 31
7.1.1 b: Home Page .................................................................................................................. 31
7.1.1 c: Home Page .................................................................................................................. 32
7.1.2: Login Page ..................................................................................................................... 32
7.1.3: Sign Up Page ................................................................................................................. 33
7.1.4: Profile Page .................................................................................................................... 33
7.1.5: Upload Foot Page........................................................................................................... 34
7.6.1: Size Detected ................................................................................................................. 34
7.1.7: Recommendation 1 ........................................................................................................ 35
7.1.8: Recommendation 2 ........................................................................................................ 35
7.1.9: Admin Page ................................................................................................................... 36
7.1.10: Logout Page ................................................................................................................. 36
 LIST OF TABLES
8.2 Test Cases ................................................................................................................................. 40
INTRODUCTION
FIND YOUR FIT 2023-2024

1. INTRODUCTION
This chapter briefs us about the description of the proposed project along with the existing
system and its limitations. The purpose, scope and applicability of project is discussed in
detail.

Fashion is a form of self-expression which mostly defines one’s identity, moods, and
opinions, which might be in any form such as clothing, footwear, accessories etc. Recent
studies show that size and fit are the factors which are most influential that drives e-
commerce for customer’s satisfaction.

When compared to traditional methods of shopping where we purchase the product in

person, one major disadvantage of using online platforms is that we cannot give immediate
feedback about how a product fits and feels. According to a study, many people might be
comfortable purchasing online products but most of them were not satisfied with the
product because of many factors like quality, fit etc., and this is a major drawback against
fashion e-commerce. To make things worse, the concept of size itself is vague: for example,
size systems of clothing may be roughly defined as small, medium and large, extra-large or
34, 36, 38 and 40 and the size systems for footwear may display the sizes according to EU,
US or UK foot size.

In fashion industry there is factor called ‘vanity sizing’ (also be termed as size inflation)
which is a phenomenon of using a ready to wear product of a nominal size but it increases
in physical size as time passes. This requires the brands to change the standardized size
specifications to target a particular set of customers. Due to the availability of so many
products that overlap size systems in the fashion industry, there is no longer a standard size
for any given product. When a product is compared with different brands and especially
when the product is related to fashion such as clothing or footwear there is no particular
method of conversion logic which is implemented to convert sizes from one to another. To
help customers find the right size, it is possible to guide them by providing size conversion
charts. Customers can convert their body measurements to the sizes of the product.
Additionally, size charts within a brand will always suffer from high variation, even if the
consumer provides their correct measurements. In this fast-moving fashion industry, the
designers of a brand try hard to compete and win against various fashion brands by offering
customers with the latest trends at reasonable and competitive prices. And further keeping
in mind the cost and design factors, and also same type of products with different attributes

PES University Department of Computer Applications 1

FIND YOUR FIT 2023-2024
like colour, material, etc often produced by different channels cause inconsistency in size
and fit factors.

There are many other factors that make size detection essential for fashion e-commerce
platforms. To develop data-driven systems and in order to provide fit and size advice to
their customers, we propose a computer vision-based content-collaborative methodology
to get personalized size and fit detection and also recommend the product of the same
resulting foot size of a desired brand. We hereby propose a logic that calculates the size of
foot and filters the footwear of a desired brand which only rely on interaction data to model
customer behaviour. The method used for this kind of measurement is based on 3D
construction which uses some materials like structured light and plantar scanner in order to
get the information about foot. And the foot information is obtained by splicing and
measuring the common parameters. Though this technique gives high accuracy, it is not
suitable to implement on a large scale as this involves working with expensive equipment
and takes a long-time consumption. So, the idea of measuring the foot based on pictures
came to light so that this process does not need any method of reconstructing the 3D shape
of foot and this process is also more convenient than that 3D measuring. The whole
measuring can be done by only clicking pictures. The logic detects edges, with more
accurate and efficient foot measurement methods. And by a particular algorithm we process
the edge information of the image, and use the traditional OpenCV algorithm to separate
the foot from the background of image to calculate the size.

1.1 PROJECT DESCRIPTION

With increasing e-commerce sales and rising trends in online shopping, our goal/motivation
is to develop a foot size detection solution for the user to improve their shopping experience
from retail stores to online platforms. Here we also recommend a product that is more
preferrable to the users along with their calculated size. And our main aim is to address the
problem of certain group of customers who are dis-satisfied with their shopping experience
as they get confused about their right size for the footwear.

As a statement made by Nike, “Three out of five people are likely to wear the wrong sized
shoe”. Our web application tries to address this problem. The web application is developed
in such way that it can replace the role of a shoe salesperson by providing the right sized
shoes to the customers. Here we measure the size of foot by using a picture captured with
a mobile phone or camera and based on the clicked picture we determine the appropriate
shoe size. The logic allows us to calculate the length, width and height of foot, making it
PES University Department of Computer Applications 2
FIND YOUR FIT 2023-2024
easier to calculate the size of the shoes in the hope of reducing the number of return products
with wrong sizes and ultimately increasing the user satisfaction.

Designed primarily for online shoes, this technology allows to accurately match different
styles of shoes which will reduce the number of shoes returned back to the retailer. There
are many concerns like either returning or exchange of the products and due to this high
rate of return, many shoe retailers make less money from online sales. Our webpage
recommends the appropriate size for shoes or type of footwear based on us oers choice.
The main focus is on user comfort on different parts of the foot, such as the toes, heels,
soles and insteps.

PURPOSE
This project aims to revolutionize fashion e-commerce by detecting foot size from user-
uploaded images. The core objective is to enhance the online shopping experience by
offering personalized footwear recommendations based on calculated foot size and user
brand preferences. With the innovative image processing techniques, the platform provides
accurate sizing, and makes the website more engaging and offers a customer-centric
shopping environment. The website seeks to improve customer satisfaction with provide a
best selection process and merge the user-specific preference with accurate foot
measurements. At the end the project aims to transform the way consumers find and buy
shoes on the internet, by maximizing the convergence of fashion and technology in the e-
commerce domain.

SCOPE
This project focuses on 2D image-based detection of human foot size. This leads to some
challenges in appropriately fitting a range of foot forms and sizes. Because each person’s
foot anatomy is unique, since it varies it might not work for everyone, which would affect
its accuracy and usefulness. It is important to consider that different users may experience
different levels of accuracy in foot sizes recognition. Thus, the research emphasizes clarity
regarding its limits in addressing the complete range of foot variability. It also gives
significant insights into automated foot sizing from uploaded images.

PES University Department of Computer Applications 3

LITERATURE
SURVEY
FIND YOUR FIT 2023-2024

2. LITERATURE SURVEY
This chapter provides an overview of the many shoe sizing systems that are in use around
the world. Some of the systems are used in the US, Great Britain, Continental Europe, and
the UK, as well as the measurement techniques unique to each of these regions. It also
presents the recently suggested BHA system, specifically used, and is designed for Indian
population. To increase shoe fitting accuracy and comfort, it also examines relevant
literature and current technological advancements that improve footwear size recognition
and recommendation.

2.1 DOMAIN SURVEY

The UK shoe size system, which is based on the measurement in barleycorns, is the most
popular in India.

UK shoe size system (measured in barleycorns):

One barleycorn equals 1/3 of an inch. So, a shoe size of 1 would be one barleycorn longer
than a shoe size 0.

US Shoe size system (measured in barleycorns):

US and UK shoe sizes are measured in barleycorns, which are roughly 1/3 in length.
However, the US starts counting from 1, so US sizes are 1 unit larger than their UK
equivalents. That is to say, a US10 is the same length as a UK9, which are both 1/3 longer
than a US9 and UK8.

Continental European Shoe Size System (Mondopoint length system):

The continental european system uses the mondopoint length system, which is based on the
length of the foot in millimeters. It is widely used in europe and other parts of the world.
The sizes are based on the actual length of the foot, making it a more precise measurement
system.

Great Britain Shoe Size System (similar to US):

The great britain shoe size system is similar to the US system, as both are derived from the
length of the brannock device. Zero-point measurement is the length of the brannock
device. It is 1/72 of an inch.

PES University Department of Computer Applications 5

FIND YOUR FIT 2023-2024
BHA (Newly Proposed Indian Shoe Sizing System)

The Central Leather Research Institute (CLRI) in collaboration with the Department for
Promotion of Industry and Internal Trade (DPIIT) led this initiative which aims to create
footwear that fits better and more comfortable for the Indian population.

This BHA emerged in a comprehensive anthropometric survey conducted involving

100000 people across 79 locations in India. The researchers by using the 3D foot scanning
Technology found that there are notable differences in the shape of the Indian forefoot (the
first half of the foot from toe onwards) from European and North American foot shapes.

There are several benefits and advantages from this proposal. It brings Improved Fit
compared to footwear based on European, US and UK moulds. It enhances the Comfort for
its users by reducing the issues like blisters, foot pain and poor blood circulation thus
bringing down the all-possible health problems in vulnerable groups such as elderly people
and diabetics. It introduces a Unified Sizing system which standardizes the shoe sizing in
India thus simplifying the process of finding the right shoe size for consumers.

The BHA has its own disadvantages since it is in its Initial phase. Implementing the new
sizing system will require a lot of time and effort from manufacturers, retailers, and
consumers to adapt to the changes. The BHA system could be expensive as well since it
involves redesigning and producing the shoes according to new BHA sizes. High
competition from established brands may resist this new sizing system.

The BHA has high future scope, as consumer become aware of this BHA system it is certain
that they will adopt this Indian sizing system. Also, other countries can adopt it for its major
advantages leading to a potential global impact.

Advances in 3D printing and custom footwear production might lead the BHA system to
offer personalized shoe solutions for its consumers.

Thus, it can be concluded that the BHA shoe-sizing system represents a significant step
forward in addressing the specific needs of the Indian population. By providing better-
fitting and more comfortable footwear, BHA has the potential to improve foot health,
enhance consumer satisfaction, and set new standards in the footwear industry.

PES University Department of Computer Applications 6

FIND YOUR FIT 2023-2024
2.2 RELATED WORKS
A relative analysis by Abdul-Saboor Sheikh [4] and his team shows in their article that, by
using a deep learning methodology they were able to combine collaborative and content-
based modeling techniques by which we could learn some useful representations of
customers for size and fit detection. Advantage of using this method is, that it is highly
scalable and also works end to end for which we do not require any knowledge in prior.
The team also proposed that by using an architecture called SFnet which gives both
flexibility and carries a good capacity in order to capture higher abstractions of fit, size
which gives related information of customers random choices. Further extension of this
work can also include detection of categorical and targets and modeling of customer
behavior with respect to size and fit. Sungkuk chun [5] and team showed in their work that
by using an autonomous geometric foot-arch analysis platform it is much more suitable to
capture the sole of the foot and produce three parameters like AI (arch index), AW (arch
width), AH (arch height). Basically, this system was proposed in order to overcome the
problem of inaccurate measurements. For this we need an RGB-D (Red Green Blue-Depth)
camera to capture a foot in a static standing pose. By applying Markov Random Fields and
based on the image segmentation methods we can detect the foot surface which is in contact
with the footplate. Sagar Arora [6] and team proposed a footwear size recommendation
system due to lack of trying out option during online shopping. Their main aim was to
provide customers with an enhanced shopping experience. The challenge encountered here
was to provide recommendations who do not have a history of any previous purchases on
online platform. A probabilistic graph was constructed which shows the brand-to-brand
relationship which address challenges like sparse signal for some new products in the
system and gives a solution for new users and the recommendation precision and coverage
was improved.

Emily Yim Lee Au and Ravindra S. Goonetilleke [7] in their study say that by using two
models they were able to predict footwear fit by considering foot-shoe dimension
difference in toe. Also, a polynomial relation was built to predict preference rating from
this footwear fit where the quality of fit will also be evaluated. Damir Omrcen and Ales
Jurca [8] share their innovative shoe size recommendation system which is based on foot
measurements that significantly increase accuracy of recommendation. Finally, they also
shared their future work on further development of this recommendation system where
instead of using foot scanners a self-foot measurement would provide a good solution.

PES University Department of Computer Applications 7

FIND YOUR FIT 2023-2024
2.3 EXISTING SYSTEM
There is various existing system which focus on Footwear size detection and
recommendation to enhance the online shopping experience and reduce the return rates.

Nike Fit: It recommends best shoe sizes when the user scan their feet using the Nike app.
It uses various concepts such as computer vision, artificial intelligence, machine learning,
data science and recommends accurate size recommendation.

True Fit: It is a platform which integrates various e-commerce sites. It also uses large
database of clothing and shoe sizes, combined with user feedback, and recommend the best
fit for the shoppers.

Volumental: It provides a 3D foot scanning technology used in shops by the shopkeepers.

It allows customers to scan their feet and get precise measurement which can be used by
the shop to recommend the right shoe size.

FeetSizr: It provides a 3D foot scanning technology and it is used by the users in their
homes through mobile apps. The app provides size recommendations based on the scanned
data and aims to improve online shoe shopping accuracy by providing a detailed fit
analysis.

DeepAI Footwear Size Recommendation System: This system uses a probabilistic

approach based on user co-purchase data to recommend footwear sizes. It uses brand to
brand relationship graph to suggest sizes across different brands.

2.4 TECHNOLOGY SURVEY

The images uploaded by users can be processed using OpenCV. The logic is built in such
a way that it begins by identifying the foot from the uploaded image, then the size is
calculated and the output is returned to the user. The user’s size is presented according to
their desired brand. By creating a user recommendation system that recommendation
system that recommends the ideal footwear for each user, the project also provides a new
feature.

2.5 FEASIBILITY STUDY

The Footwear Size Detection and Recommendation System seeks to improve the online
buying experience; hence a feasibility study is necessary to assess the system’s practicality.
The technological, operational, and economic feasibility is assessing in this study.

PES University Department of Computer Applications 8

FIND YOUR FIT 2023-2024
Technological Feasibility: OpenCV is notably used for edge detection and image
segmentation in this project. It will accept the user uploaded image. By using these
methods, the system will be able to give the length, height and width of a foot from images
that are uploaded by the users. The algorithms will evaluate the edge data, distinguish the
foot from the background and provides necessary measurements. The models based on
machine learning will suggest suitable footwear sizes and guarantees its user an accurate
fit and recommendations.
The use of mobile phone cameras for capturing images makes it easy for the users to upload
images of the foot. The system uses known frameworks like OpenCV and machine learning
libraries such as TensorFlow or PyTorch.
Operational Feasibility: The proposed project involves users capturing images of their
feet using their smartphones, which are accepted by the website, later used to determine the
sizes and display data. It also comes with User Instructions that give clear guidelines on
how to take good pictures of the foot. It has image processing which is mainly done using
advanced algorithms and identifies edges of the image. Using machine learning algorithms
it analyses the measurements, precise suggestions for shoe sizes of various brands. Detailed
instructions and user-friendly interfaces can help to alleviate operational issues like image
quality variances and user compliance. It is also critical to guarantee data privacy and
handle user foot photos securely. These elements can be addressed so that the system can
function successfully and efficiently, making the project operationally feasible.
Economical Feasibility: It mainly focuses on budget feasibility of the users as well as the
developers. It includes development costs, operational costs, Marketing costs, Maintenance
charge etc. The project involves development costs for the initial development of the web
application and image processing algorithms. It also includes the costs of integration of
machine learning models.
Hosting the server and maintaining the web application also should be considered in the
budget. To make the world know about the website, promoting it is important. The cost
involved in this is called marketing costs.
This is a viable project with high feasibility across technological, operational, and
economic dimensions. The use of established technologies like OpenCV and machine
learning frameworks ensures technical success. A well-designed operational workflow
supports practical implementation, and various revenue models provide financial
sustainability. This feasibility study indicates that the project has strong potential to
enhance the online shopping experience and generate significant value in the fashion e-
commerce industry.
PES University Department of Computer Applications 9
HARDWARE AND
SOFTWARE
REQUIREMENTS
FIND YOUR FIT 2023-2024

3. HARDWARE AND SOFTWARE REQUIREMENTS

This chapter discuses the essential hardware and software requirements along with it’s
versions for the development and deployment of the Footwear Size Detection and
Recommendation System.

3.1 HARDWARE REQUIREMENTS

 Processor: Intel(R) Core (TM) i3-6006U CPU @ 2.00GHz, 2000 MHz, 2 Core(s),
4 Logical Processor(s) and higher.
 System Type: x64-based PC and higher.
 RAM: Minimum of 8GB is recommended.
 System Memory: 512 GB SSD or 1 TB HDD.

3.2 SOFTWARE REQUIREMENTS

 Django Framework: Version 4.0

 Python: Version: 3.10
 SQLite: Version: 3.45

PES University Department of Computer Applications 11

 SOFTWARE
REQUIREMENTS
SPECIFICATION
FIND YOUR FIT 2023-2024

4. SOFTWARE REQUIREMENTS SPECIFICATION

This chapter outlines the functional and non-functional requirements for a website that
identifies foot size from uploaded images. The website aims to give precise foot size
measurements and a smooth user experience, reaching a wide range of users, including
consumers, retailers, and researchers.

4.1 USERS
With the foot size identification, the website is designed to accommodate a wide range of
users with various needs and interest. These users include consumers and online shoppers
seeking accurate measurements, shoe manufacturers wishing to integrate foot size detection
technology into their manufacturing processes, and retailers and e-commerce sites hoping
to enhance customer experience and reduce return rates. Researchers and developers
studying or developing foot size detection techniques and applications in fields such as
computer vision and artificial intelligence may also find the website useful. Lastly, casual
visitors who are interested in foot size detection technology or machine learning
applications in everyday life may find the website useful. The website seeks to provide
precise outcomes and is user-friendly to the users.

4.2 FUNCTIONAL REQUIREMENTS

OpenCV integration: This Creates the logic to recognize and extract the foot image
uploaded by the user using OpenCV.

Foot Size Detection: The system can precisely determine the size of the foot by using the
foot image by the user. This size detection calculates the foot size and displays the size in
centimetres, euros, pounds, and US.

Recommendation System: A few variables, including brand preferences, style

preferences, is considered when making the recommendation system.

User Interface Integration: Provide an easy-to-use interface and makes it possible for
users to upload pictures of their feet. The system provides guidelines for capturing images
gives an easy-to-use interface which identifies the foot size and recommends brands to the
user.

PES University Department of Computer Applications 13

FIND YOUR FIT 2023-2024
Feedback Mechanism: Provide a means for consumers to provide comments regarding
how well the suggested footwear fits. Future suggestions will be more accurate because to
the continual improvement of the recommendation algorithm made possible by this
feedback.

4.3 NON-FUNCTIONAL REQUIREMENTS

Performance: The minimum amount of response time is required. To improve
performance, the uploaded photo must be processed as quickly as possible.
Security: There should be no data leakage from the user-uploaded photographs in the
database.
Accuracy: The system should recommend shoe size with high accuracy to minimize user
dissatisfaction and returns.
Usability: The web application should be user-friendly and easy to navigate. Uploading
pictures and selecting brands should be intuitive.
Scalability: The system should be able to handle many users and image uploads.
Availability: The web application should be available for use most of the time with
minimal downtime.
Compatibility: The web application should function properly on different web browsers
and devices (phones, tablets, laptops).

PES University Department of Computer Applications 14

SYSTEM DESIGN
FIND YOUR FIT 2023-2024

5. SYSTEM DESIGN
This chapter describes how the system is designed using two diagrams namely Architecture
diagram and Context flow diagram. This shoe size recommendation website utilizes a 3-
tier client-server architecture. This document details the architecture and interaction
between the user interface, processing logic, and data storage components.

5.1 ARCHITECTURE DIAGRAM

The project follows a Client-Server Architecture where the client sends the request and the
server accepts the request and processes it. At last, the server delivers the response back to
the client. The project is implemented in a 3 Tier architecture where the user interacts with
the web browser namely the Presentation Tier and the request sent by the user is taken into
the Web Server where the logics are applied hence it is named the Logic Tier. The data for
these queries are fetched from the Database where csv files are used for storing the data
and these are saved in tire named Data Tier. Hence in this way the project follows a 3 Tier
Architecture and the same is depicted in the figure below.

Figure 5.1: Architecture Diagram

PES University Department of Computer Applications 16

FIND YOUR FIT 2023-2024

5.2 BLOCK DIAGRAM / CONTEXT FLOW DIAGRAM

The context diagram illustrates about the shoe size recommendation system. It shows that
the system interacts with two external entities: the user and a database. The user uploads a
picture of their foot which is then sent to the system. The system interacts with a database,
likely containing shoe size charts for various brands, to retrieve relevant information.
Internally, the system analyses the uploaded picture using computer vision to measure the
user's foot. Finally, the system recommends a shoe size based on the foot measurements
and retrieved size information and displays this recommendation to the user.

Figure 5.2: Context Diagram

PES University Department of Computer Applications 17

DETAILED DESIGN
FIND YOUR FIT 2023-2024

6. DETAILED DESIGN
In this chapter the detailed design of the shoe size detection and recommendation system
explores the methodology used for image pre-processing, foot detection through
segmentation and edge detection, and size calculation. Additionally, the design
incorporates a user-friendly recommendation system for footwear selection.

6.1 PROCESS FLOW DIAGRAM WITH DESIGN METHODOLOGY

The raw image is first converted into the HSV color space in the image processing pipeline.
Gaussian blur is used to eliminate noise of the raw image and improve the quality of the
image. K-means Clustering is used to group pixels with similar values for color-based
segmentation. Then, edge detection is used to identify boundaries in the clustered picture.
The edges that have been discovered are used to identify contours, which helps distinguish
between different shapes such as rectangle A4sheet, and foot structure. Bounding boxes are
created around the recognized contours to provide exact height and width measurements to
find the size of a paper and feet within the image.

Figure 6.1: Process Flow Diagram

PES University Department of Computer Applications 19

FIND YOUR FIT 2023-2024
6.2 METHODOLOGY
The image uploaded by the user (Fig 6.2.1) is pre-processed into Hue Saturation Value
(HSV) format for better colour classification. Gaussian Blur function is used to smoothen
the image uploaded (Fig 6.2.2). Then the image is normalised by dividing it with the
maximum HSV value, i.e., 225. This pre-processing will help machine to identify the
different features in the image. Then the image is converted into 2D and fed into KMeans
Clustering algorithm. This helps to cluster the image into two parts, background of image
and foot of user along with A4 sheet (Fig 6.2.3). The clustered image is fed into Edge
Detection function to calculate the edges of the image. In the function, a filter matrix
namely “canny” is multiplied with clustered image pixel to detect the edges. Dilate and
Erode methods are used to remove the noise from the clustered image (Fig 6.2.4). Bounding
box is used to detect the boundary of the foot and used to extract counters for different
colour patterns. In Draw counters, CV2.rectangle method is implemented to draw rectangle
between foot and page (Fig 6.2.5). Then, the counters are cropped from the original image.
This cropped image is overlayed with the original image to check for differences and noise
present in the cropped image (Fig 6.2.6).

The overlayed image is fed into Edge Detection function again to crop the foot alone
excluding the paper (Fig 6.2.7). Once the foot and paper edges are extracted, Fbound Rect
function is applied on foot and paper to calculate the height and width of images. Based
on height and width, distance metric is used to compute the size of the foot.

Figure 6.2.1: Input Image

PES University Department of Computer Applications 20

FIND YOUR FIT 2023-2024

Figure 6.2.2: Pre-Processing

Figure 6.2.3: K-Means Clustering

Figure 6.2.4: Edge Detection

PES University Department of Computer Applications 21

FIND YOUR FIT 2023-2024

Figure 6.2.5: Contours Image

Figure 6.2.6: Cropped Image

Figure 6.2.7: Overlay Image

PES University Department of Computer Applications 22

FIND YOUR FIT 2023-2024
RECOMMENDATION
 The website has a simple user-friendly Recommendation system. It will simplify
the footwear selection process.
 The User can select their gender, preferred brand, and footwear type in dropdown
menus. And user can specify their footwear size, resulting in a link that directs the
users to a partner website.
 This feature saves time and effort in the search process making the website as a
User-friendly website.

PES University Department of Computer Applications 23

IMPLEMENTATION
FIND YOUR FIT 2023-2024

7. IMPLEMENTATION
This chapter consists of Pseudo code and provides a visual walkthrough of the user
experience, allowing you to see the website's functionalities in action. It includes Home
page, Login page, profile Page, sign up page, Profile page, size Detection pages and
Recommendation page

7.1 PSEUDOCODE

def size(request):
if request.method == 'POST':
p_form =
footimage(request.POST,request.FILES,instance=request.user.footimage)
if p_form.is_valid():
p_form.save()
messages.success(request, 'Here are your results')
a = p_form.cleaned_data.get("image")
oimg = p_form
preprocessedOimg = preprocess(a)
clusteredImg = kMeans_cluster(preprocessedOimg)
edgedImg = edgeDetection(clusteredImg)
boundRect, contours, contours_poly, img = getBoundingBox(edgedImg)
pdraw = drawCnt(boundRect[1], contours, contours_poly, img)
croppedImg, pcropedImg = cropOrig(boundRect[1], clusteredImg)
newImg = overlayImage(croppedImg, pcropedImg)
fedged = edgeDetection(newImg)
fboundRect, fcnt, fcntpoly, fimg = getBoundingBox(fedged)
fdraw = drawCnt(fboundRect[2], fcnt, fcntpoly, fimg)
output = calcFeetSize(pcropedImg, fboundRect)/10

output_m = output
output_w = output

if output_m < 23.5:

UK_M = "Invaild Image"
US_M = "Invaild Image"

PES University Department of Computer Applications 25

FIND YOUR FIT 2023-2024
Euro_M = "Invaild Image"
elif output_m >= 23.5 and output_m < 24.1:
UK_M = 5.5
US_M = 6
Euro_M = 39
elif output_m >= 24.1 and output_m < 24.4:
UK_M = 6
US_M = 6.5
Euro_M = 39
elif output_m >= 24.4 and output_m < 24.8:
UK_M = 6.5
US_M = 7
Euro_M = 40
elif output_m >= 24.8 and output_m < 25.4:
UK_M = 7
US_M = 7.5
Euro_M = "40 - 41"
elif output_m >= 25.4 and output_m < 25.7:
UK_M = 7.5
US_M = 6
Euro_M = 41
elif output_m >= 25.7 and output_m < 26:
UK_M = 8
US_M = 8.5
Euro_M = "41 - 42"
elif output_m >= 26 and output_m < 26.7:
UK_M = 8.5
US_M = 9
Euro_M = 42
elif output_m >= 26.7 and output_m < 27:
UK_M = 9
US_M = 9.5
Euro_M = "42 - 43"
elif output_m >= 27 and output_m < 27.3:
UK_M = 9.5
PES University Department of Computer Applications 26
FIND YOUR FIT 2023-2024
US_M = 10
Euro_M = 43
elif output_m >= 27.3 and output_m < 27.9:
UK_M = 10
US_M = 10.5
Euro_M = "43 - 44"
elif output_m >= 27.9 and output_m < 28.3:
UK_M = 10.5
US_M = 11
Euro_M = 44
elif output_m >= 28.3 and output_m < 28.6:
UK_M = 11
US_M = 11.5
Euro_M = "44 - 45"
elif output_m >= 28.6 and output_m < 29.4:
UK_M = 11.5
US_M = 12
Euro_M = 45
elif output_m >= 29.4 and output_m < 30.2:
UK_M = 12.5
US_M = 13
Euro_M = 46
elif output_m >= 30.2 and output_m < 31:
UK_M = 13.5
US_M = 14
Euro_M = 47
elif output_m >= 31 and output_m < 31.8:
UK_M = 14.5
US_M = 15
Euro_M = 48
elif output_m >= 31.8:
UK_M = 15.5
US_M = 16
Euro_M = 49

PES University Department of Computer Applications 27

FIND YOUR FIT 2023-2024
print("UK Size: ",UK_M)
print("US Size: ",US_M)
print("Euro Size: ",Euro_M)

if output_w < 20.8:

UK_W = "Invalid Image"
US_W = "Invalid Image"
Euro_W = "Invalid Image"
elif output_w >= 20.8 and output_w < 21.3:
UK_W = 2
US_W = 4
Euro_W = 35
elif output_w >= 21.3 and output_w < 21.6:
UK_W = 2.5
US_W = 4.5
Euro_W = 35
elif output_w >= 21.6 and output_w < 22.2:
UK_W = 3
US_W = 5
Euro_W = "35 - 36"
elif output_w >= 22.2 and output_w < 22.5:
UK_W = 3.5
US_W = 5.5
Euro_W = 36
elif output_w >= 22.5 and output_w < 23:
UK_W = 4
US_W = 6
Euro_W = "36 - 37"
elif output_w >= 23 and output_w < 23.5:
UK_W = 4.5
US_W = 6.5
Euro_W = 37
elif output_w >= 23.5 and output_w < 23.8:
UK_W = 5
US_W = 7
PES University Department of Computer Applications 28
FIND YOUR FIT 2023-2024
Euro_W = "37 - 38"
elif output_w >= 23.8 and output_w < 24.1:
UK_W = 5.5
US_W = 7.5
Euro_W = 38
elif output_w >= 24.1 and output_w < 24.6:
UK_W = 6
US_W = 8
Euro_W = "38 - 39"
elif output_w >= 24.6 and output_w < 25.1:
UK_W = 6.5
US_W = 8.5
Euro_W = 39
elif output_w >= 25.1 and output_w < 25.4:
UK_W = 7
US_W = 9
Euro_W = "39 - 40"
elif output_w >= 25.4 and output_w < 25.9:
UK_W = 7.5
US_W = 9.5
Euro_W = 40
elif output_w >= 25.9 and output_w < 26.2:
UK_W = 8
US_W = 10
Euro_W = "40 - 41"
elif output_w >= 26.2 and output_w < 26.7:
UK_W = 8.5
US_W = 10.5
Euro_W = 41
elif output_w >= 26.7 and output_w < 27.1:
UK_W = 9
US_W = 11
Euro_W = "41 - 42"
elif output_w >= 27.1 and output_w < 27.6:
UK_W = 9.5
PES University Department of Computer Applications 29
FIND YOUR FIT 2023-2024
US_W = 11.5
Euro_W = 42
elif output_w >= 27.6:
UK_W = 10
US_W = 12
Euro_W = "42 - 43"

print("US Size: ",US_W)

print("UK Size: ",UK_W)
print("Euro Size: ",Euro_W)

output_m = round(output_m, 2)
output_w = round(output_w, 2)

sizes = [
{
'Gender' : "Male",
'UK_Size': UK_M,
'US_Size': US_M,
'Euro_Size': Euro_M,
'Centimeters' : output_m
},
{
'Gender' : "Female",
'UK_Size': UK_W,
'US_Size': US_W,
'Euro_Size': Euro_W,
'Centimeters': output_w
}
]
context = {'sizes':sizes}
return render(request,'Prediction/output.html',context)
else:
p_form = footimage(instance=request.user.profile)

PES University Department of Computer Applications 30

FIND YOUR FIT 2023-2024
context = {
'p_form': p_form
}

return render(request, 'Prediction/size.html', context)

7.2 SCREENSHOT

Figure 7.1.1 a: Home Page

The home page includes several navigation bars such as Home, About, Profile, Logout,
Know Your Size, and Recommendation sections. It also has a "Contact Us" button, inviting
users to reach out for more information or assistance

Figure 7.1.1 b: Home Page

PES University Department of Computer Applications 31
FIND YOUR FIT 2023-2024

Figure 7.1.1 c: Home Page

Figure 7.1.2: Login Page

The page is titled "LOG IN" and features fields for users to enter their username and
password, along with a "Login" button to submit their credentials. Below the login form,
there's a prompt that says "Need An Account? Sign In," directing new users to create an
account if they do not have one.

PES University Department of Computer Applications 32

FIND YOUR FIT 2023-2024

Figure 7.1.3: Sign Up Page

The Sign-Up page is designed for new users to create an account. t includes form fields
for entering a username, email, password, and password confirmation. The username
field allows up to 150 characters, including letters, digits, and certain special characters.
The password field specifies requirements: it must be at least 8 characters long, not
similar to other personal information, not a commonly used password, and not entirely
numeric.

Figure 7.1.4: Profile Page

The Profile page is designed to maintain the information of the user. This provides
personalization. It saves the name, email id and image of the User. The user can update
the information anytime in the website.

PES University Department of Computer Applications 33

FIND YOUR FIT 2023-2024

Figure 7.1.5: Upload Foot Page

The Know your Size page will have an option for the user to upload his/her Foot image.
The image which is to be uploaded by the user should satisfy several conditions. The image
should have its foot at the center touching at one edge only, Image must be captured from
the top angle, Paper and background must be of different colours. It also has a Submit
Button. The user can upload the image and submit it. It displays a alert message requesting
the user to wait for some time.

Figure 7.6.1: Size Detected

The image displays a size chart for shoes, indicating different sizes based on gender.
For males, a US size 7 corresponds to UK size 6.5, Euro size 40, and 24.72 centimetres.
For females, a US size 8.5 is equivalent to UK size 6.5, Euro size 39, and also
24.72 centimetres.
PES University Department of Computer Applications 34
FIND YOUR FIT 2023-2024

Figure 7.1.7: Recommendation 1

The image shows a shoe recommendation interface where users can select their gender,
preferred brand, type of wear, and shoe size. In this instance, the selected options are Male,
Bata brand, Sports wear, and size 7, with a submission button leading to a recommendation
link for the specified criteria.

Figure 7.1.8: Recommendation 2

The image shows a sandals recommendation interface where users can select their gender,
preferred brand, type of wear, and shoe size. In this instance, the selected options are
Female, Puma brand, sandals, and size 9, with a submission button leading to a
recommendation link for the specified criteria.

PES University Department of Computer Applications 35

FIND YOUR FIT 2023-2024

Figure 7.1.9: Admin Page

The image describes the Django admin page. Here all the admin privileges can be seen.
The admin has authority to create new group, users. Admin can delete them as well. The
recent actions of the admin will be displayed under recent actions. The red cross mark
means the admin has removed the user and green tick mark means the admin has added the
user.

Figure 7.1.10: Logout Page

The Logout page will logout the user from the website FIND YOUR FIT. Also it has a
option which provides the user a chance to login again if the user wish to.

PES University Department of Computer Applications 36

SOFTWARE
TESTING
FIND YOUR FIT 2023-2024

8. SOFTWARE TESTING
This chapter discusses about various test cases and the approaches used for testing the
model.

8.1 TEST CASES

Test Case ID: Test Plan Expected Actual Result Pass / Fail
Result
TC001 Registration Sign up Same as Pass
successful expected
TC002 Sign Up with User with Same as Pass
same name name already expected
exists
TC003 Password similar The password Same as Pass
to user name is too similar expected
to the
username.
TC004 Password less This password Same as Pass
than 8 characters is too short. It expected
must contain
at least 8
characters
TC005 Password is This password Same as Pass
numeric is entirely expected
numeric.
TC006 Different The two Same as Pass
passwords during password expected
confirmation fields didn’t
match.
TC007 Login with Please enter a Same as Pass
unregistered user correct expected
username and
password.
Note that both

PES University Department of Computer Applications 38

FIND YOUR FIT 2023-2024
fields may be
case-sensitive.
TC008 Updating user Your account Same as Pass
Information has been expected
updated!
TC009 User uploads foot Please wait as Same as Pass
image in correct it may take expected
format some time to
progress
TC010 User uploads foot Thank you for Same as Pass
image in correct your patience. expected
format Here are your
results
TC011 User uploads Invalid Image Same as Pass
wrong Image expected
TC012 Recommendation Link of the Same as Pass
of product website pops expected
selected by user up in alert box
(If exists) and same
should be
displayed in
text.
TC013 Recommendation Product out of Same as Pass
of product stock expected
selected by user
(If it doesn’t
exist)
TC014 Logout from the You have Same as Pass
website been logged expected
out, Log In
Again
TC015 Authorized user Login Same as Pass
logins to admin successful expected
page

PES University Department of Computer Applications 39

FIND YOUR FIT 2023-2024
TC016 Unauthorized Please enter Same as Pass
user logins to the correct expected
admin page username and
password for a
staff account.
Note that both
fields may be
case-sensitive.
Table 18.2 Test Cases

8.2 MODEL TESTING:

High Accuracy Rate: This foot size detection model, a computer vision-based approach
demonstrates impressive accuracy, accurately identifying foot sizes in 8 out of 10 input
images.

Quantified Accuracy: This results in an 80% accuracy rate, highlighting the model's
effectiveness in detecting foot measurements from user-provided images.

Further Improvement: While the model excels in most cases, refinement and
optimization efforts may be needed to address the remaining 20% of inaccuracies.

Continuous Enhancement: There is a need of Consistent evaluation and adjustment of the

model's algorithms to improve its performance and reliability. Thus ensuring precise foot
size predictions for an enhanced online shopping experience.

PES University Department of Computer Applications 40

CONCLUSION
FIND YOUR FIT 2023-2024

9. CONCLUSION
In summary, our project offers several benefits that enhance the user experience. The
website's simplicity ensures ease of use, with updates reflecting instantly across the
platform. The conversion of foot sizes into various measurements like Euro, UK, and
US sizes provides users with valuable information, making their shopping experience
more convenient. Also, the secure storage of user-uploaded pictures ensures data
privacy and builds trust among the users. Additionally, the integration of foot detection
with a recommendation system adds another layer of personalization, catering to
individual preferences. However, certain limitations, such as specific image capture
requirements and the need for repetitive uploads are some of the present challenges
that need addressing. By overcoming these hurdles, we can further improve our project
and make it even more attractive to users seeking accurate foot size detection and
seamless online shopping experiences.

PES University Department of Computer Applications 42

 FUTURE
ENHANCEMENT
FIND YOUR FIT 2023-2024

10. FUTURE ENHANCEMENT

The project can be extended with much more advance processing speed. Currently
the processing of the image takes time to fetch the results which can be replaced
with advanced logics to fetch results faster. The application currently has only one
option to upload the image which can be extended with a QR code system where
the user scans the QR and clicks the image which can automatically be reflected on
the website. The image currently requires the user to click picture with A4 at
background which can be generalised without it. The webpage can be extended to
different language-based use so that many users can use the application. The
Recommendation system can be improved such that as soon as the user selects the
category, the webpage of that category opens rather than displaying the link of the
website.

PES University Department of Computer Applications 44

APPENDIX A
FIND YOUR FIT 2023-2024

11. APPENDIX A
PAPER REFERENCES
[1] A.-S. Sheikh et al., “A Deep Learning System for Predicting Size and Fit
in Fashion E-Commerce,” Proceedings of the 13th ACM Conference on
Recommender Systems, pp. 110–118, Sep. 2019, doi:
10.1145/3298689.3347006.

[2] Gunseli Yalcinkaya, “Nike app uses AR and AI to scan feet for perfect
fit,” Dezeen, May 09, 2019. https://www.dezeen.com/2019/05/09/nike-fit-app-
ar-ai-trainers/

[3] Wevolver, “Computer vision for virtual footwear fitting and shoe size
suggestion,” 1651. https://www.wevolver.com/article/computer-vision-for-
virtual-footwear-fitting-and-shoe-size-suggestion (accessed Jul. 29, 2022).

[4] A.-S. Sheikh et al., “A deep learning system for predicting size and fit in fashion
e-commerce,” in Proceedings of the 13th ACM Conference on Recommender
Systems, 2019.

[5] Chun, S. et al. (2017) “A foot-arch parameter measurement system using a

RGB-D camera,” Sensors, 17(8), p. 1796. Available at:
https://doi.org/10.3390/s17081796.

[6] S. Singh, G. M. Abdulla, S. Borar, and S. Arora, “Footwear size

recommendation system,” arXiv [cs.IR], 2018.

[7] E. Y. L. Au and R. S. Goonetilleke, “A psychophysical model for predicting

footwear fit,” in 2009 IEEE International Conference on Virtual Environments,
Human-Computer Interfaces and Measurements Systems, 2009, pp. 202–205.

[8] D. Omrcen and A. Jurca, “Shoe size recommendation system based on shoe
inner dimension measurement,” in Proceedings of the 2nd International Conference
on 3D Body Scanning Technologies, Lugano, Switzerland, 25-26 October 2011,
2011.

PES University Department of Computer Applications 46

FIND YOUR FIT 2023-2024

WEB REFERENCES
[9] Foot-scanning-app (no date) Find Me A Shoe. Available at:
https://findmeashoe.com/foot-scanning-
app/#:~:text=Findmeashoe%20foot%20scanner%20app%20allows,ordering%20th
e%20best%20fitting%20shoe. (Accessed: 03 June 2024).

[10] Connecting Foot and shoe fitting together (no date) Connecting Foot and Shoe
Fitting Together | Shoefitr.io. Available at: https://www.shoefitr.io/ (Accessed: 03
June 2024).

[11] Python django tutorial: Full-featured web app part 1 - getting started (2018)
YouTube. Available at: https://www.youtube.com/watch?v=UmljXZIypDc
(Accessed: 03 June 2024).

[12] Python django tutorial for beginners (2021) YouTube. Available at:
https://www.youtube.com/watch?v=rHux0gMZ3Eg (Accessed: 03 June 2024).

[13] Find your shoe size (2020) YouTube. Available at:

https://www.youtube.com/watch?v=-55OxDGmTL8 (Accessed: 03 June 2024).

[14] How to measure your shoe size | daraz nepal | (2018) YouTube. Available at:
https://www.youtube.com/watch?v=kPQ2s_NeTMU (Accessed: 03 June 2024).

[15] Django project introduction: Django project tutorial [1] (2020) YouTube.
Available at: https://www.youtube.com/watch?v=gAI218HSK8s (Accessed: 03
June 2024).

PES University Department of Computer Applications 47

APPENDIX B
FIND YOUR FIT 2023-2024

12. PLAGIARIASM REPORT

PES University Department of Computer Applications 49

FIND YOUR FIT 2023-2024

PES University Department of Computer Applications 50

APPENDIX C
FIND YOUR FIT 2023-2024

13. POSTER

PES University Department of Computer Applications 52