SmartCart

Capstone Project Report

MID SEMESTER EVALUATION
Submitted by:

102003452 Kunwar Apoorvaditya

102003037 Himangi Sharma

102053010 Vyom Kapur

102053005 Aditya Kalhan

102017166 Pratham Aggarwal

BE Third Year- COE/ CSE

CPG No. _28__

Under the Mentorship of

Dr. Ravinder Kumar

Associate Professor

Computer Science and Engineering Department

Thapar Institute of Engineering and Technology, Patiala

August 2023
 ABSTRACT

The SmartCart project presents an innovative solution aimed at revolutionizing the shopping
experience within convenience stores. This report encapsulates the project's comprehensive
approach, combining advanced object recognition technology, machine learning algorithms,
and a web application to streamline shopping processes. The SmartCart system automatically
identifies items placed in the cart, expediting checkout and minimizing manual intervention.
The companion web application facilitates cashless payments, personalized recommendations,
and efficient store navigation, enhancing user engagement. A weight-based system prevents
fraudulent activities, ensuring accurate billing and security. This report delves into the technical
architecture, hardware and software requirements, performance benchmarks, risk analysis, and
regulatory considerations of the SmartCart project. Through careful integration of cutting-edge
technologies and meticulous planning, the project strives to redefine convenience stores'
operational efficiency and customer satisfaction.

2
 DECLARATION

We hereby declare that the design principles and working prototype model of the project
entitled SmartCart is an authentic record of our own work carried out in the Computer Science
and Engineering Department, TIET, Patiala, under the guidance of Dr. Ravinder Kumar and
during 6th semester (2023).

Date:

Roll No. Name Signature

102003452 Kunwar Apoorvaditya

102003037 Himangi Sharma

102053010 Vyom Kapur

102053005 Aditya Kalhan

102017166 Pratham Aggarwal

Counter Signed By:

Faculty Mentor:

Dr. _Ravinder Kumar

Designation: Assistant Professor

CSED,

TIET, Patiala

3
 ACKNOWLEDGEMENT

We would like to express our thanks to our mentor Dr. Ravinder Kumar. He has been of great
help in our venture, and an indispensable resource of technical knowledge. He is truly an
amazing mentor to have.

We are also thankful to Dr. Shalini Batra, Head, Computer Science and Engineering
Department, entire faculty and staff of Computer Science and Engineering Department, and
our friends who devoted their valuable time and helped us in all possible ways towards
successful completion of this project. We thank all those who have contributed either directly
or indirectly towards this project.

Lastly, we would also like to thank our families for their unyielding love and encouragement.
They always wanted the best for us and we admire their determination and sacrifice.

Date:

Roll No. Name Signature

102003452 Kunwar Apoorvaditya

102003037 Himangi Sharma

102053010 Vyom Kapur

102053005 Aditya Kalhan

102017166 Pratham Aggarwal

4
 TABLE OF CONTENTS

ABSTRACT……………………………………………………………………………….….2
DECLARATION………………………………………………………………………….….3
ACKNOWLEDGEMENT……………………………………………………..………….....4
LIST OF FIGURES……………………………...…………………………………….….....7
LIST OF TABLES…………………………………………………………………………...8
LIST OF ABBREVIATIONS…………………………………………………………….....9

CHAPTER……………………………………………………………………………PageNo.
1. Introduction 10
1.1 Project Overview
1.2 Need Analysis (1 page- mentioning significance of this work)
1.3 Research Gaps (Identify and explain at least five research gaps with references)
1.4 Problem Definition and Scope
1.5 Assumptions and Constraints
1.6 Standards
1.7 Approved Objectives
1.8 Methodology
1.9 Project Outcomes and Deliverables
1.10 Novelty of Work
2. Requirement Analysis 23
2.1 Literature Survey
2.1.1 Theory Associated With Problem Area
2.1.2 Existing Systems and Solutions
2.1.3 Research Findings for Existing Literature
2.1.4 Problem Identified
2.1.5 Survey of Tools and Technologies Used
2.2 Software Requirement Specification
2.2.1 Introduction
[Link] Purpose
[Link] Intended Audience and Reading Suggestions
[Link] Project Scope
2.2.2 Overall Description
[Link] Product Perspective
[Link] Product Features
2.2.3 External Interface Requirements
[Link] User Interfaces
[Link] Hardware Interfaces
[Link] Software Interfaces
2.2.4 Other Non-functional Requirements
[Link] Performance Requirements
[Link] Safety Requirements

5
 [Link] Security Requirements
2.3 Cost Analysis
2.4 Risk Analysis
3. Methodology Adopted 34
3.1 Investigative
3.2 Proposed Solution (2-3 pages)
3.3 Work Breakdown Structure
3.4 Tools and Technology
4. Design Specifications 43
4.1 System Architecture
4.2 Design Level Diagrams
4.3 User Interface Diagrams
4.4 Snapshots of Working Prototype
5. Conclusions and Future Scope 46
5.1 Work Accomplished (Discussion w.r.t. the approved objectives)
5.2 Conclusions
5.3 Environmental (/ Economic/ Social) Benefits
5.4 Future Work Plan
APPENDIX A: Reference 48
APPENDIX B: Plagiarism Report 50

6
 LIST OF TABLES

Table No. Caption Page no.

1 List of Assumptions 15

2 List of Constraints 16

3 Literature Survey Papers 23

7
 LIST OF FIGURES

Figure No. Caption Page No.

1 Cost Analysis 32

2 System Architecture Diagram 43

3 Design Level Diagram 43

4 User Interface Diagram 44

5 Snapshot of the prototype detecting the object 44

Snapshot of the prototype adding the object in the bill 45

6
after passing the line
Snapshot of the Database after the object is added to the 45
7
bill

8
 LIST OF ABBREVIATIONS

9
 INTRODUCTION

1.1 Project Overview

In today's fast-paced world, innovation continues to transform every facet of our lives, and
shopping is no exception. The advent of e-commerce has reshaped how we purchase goods,
but the traditional brick-and-mortar shopping experience has remained largely unchanged.
Enter the SmartCart, a revolutionary concept poised to streamline shopping processes, redefine
convenience, and reshape customer interactions within physical retail spaces.

The SmartCart project centres on the creation of an intelligent shopping cart that brings
together various technologies to enhance the shopping experience. The cart aims to alleviate
some of the common pain points associated with in-store shopping, including long billing
queues, manual item scanning, and a lack of personalized recommendations.

A cornerstone of the SmartCart's functionality lies in its ability to automate the billing process.
Object recognition technology, a subset of artificial intelligence, is employed to effortlessly
identify items placed within the cart. By leveraging sophisticated algorithms and cameras, the
system recognizes products based on their visual features, such as shape, color, and packaging.

This approach eliminates the need for manual barcode scanning, expediting the checkout
process and reducing customer wait times. The system's accuracy and speed are pivotal in
ensuring a smooth and efficient shopping experience.

To mitigate the risk of fraudulent activities like shoplifting, the project incorporates a weight-
based system. The cart's total weight is continually monitored and compared against the
theoretical weight of all the items placed within. This approach serves as a safeguard against
unauthorized removal of items from the cart, enhancing the integrity of the shopping process.

While the weight-based system is a creative and practical solution, it is important to address
challenges related to items with irregular shapes or packaging, as well as variations in product
weights due to factors like moisture or packaging materials.

10
The SmartCart's companion web application further elevates the shopping experience.
Designed with user-centricity in mind, the app serves as a digital hub for customers, offering
features such as on-the-spot cashless payments, personalized product recommendations and a
shopping list tracker.

With a user-friendly interface, the app allows customers to seamlessly connect with their carts,
view real-time item updates, and explore product suggestions tailored to their preferences.
Cashless payments are enabled through secure encryption protocols, ensuring the safety of
sensitive payment information.

Machine learning algorithms play a pivotal role in enhancing customer engagement. The
SmartCart's recommendation system leverages historical shopping data, current selections, and
even trends to offer relevant product suggestions. This personalized touch not only aids
customers in discovering new items but also contributes to increasing the average transaction
value for store owners.

Collaborative filtering, content-based filtering, and hybrid approaches are among the
techniques utilized to create a dynamic and accurate recommendation engine. Balancing
accuracy with avoiding "filter bubbles" is a key consideration in offering diverse and appealing
product suggestions.

Beyond its customer-centric benefits, the SmartCart project has far-reaching implications for
convenience store owners. By automating the billing process, the need for dedicated checkout
counters is significantly reduced. This spatial optimization allows store owners to reallocate
floor space for additional product displays or customer seating areas, enhancing the overall
shopping environment.

Moreover, the project aligns with the goal of cost reduction for store owners. Eliminating the
need for manual item scanning minimizes the reliance on traditional barcode scanners and
reduces the associated maintenance costs. In turn, this contributes to potential savings in
employee salaries dedicated to checkout operations.

Ensuring the successful adoption of the SmartCart requires effective user education. Store
employees and customers alike should receive clear instructions on operating the SmartCart
11
and utilizing its companion app. Creating user guides, video tutorials, and in-store support
channels can facilitate a smooth transition to the new shopping paradigm.

Additionally, accessibility considerations are paramount. The project should aim to

accommodate individuals with disabilities by incorporating features such as voice-guided
navigation and compatibility with screen readers for visually impaired users.

To validate the effectiveness and reliability of the SmartCart, comprehensive real-world testing
is imperative. Controlled testing environments within convenience stores will provide valuable
insights into the system's performance, accuracy of object recognition, and the user experience.
Feedback from customers and store employees during these trials will inform iterative
improvements to fine-tune the technology.

As the project involves sensitive customer data, regulatory compliance and data privacy are
non-negotiable aspects. Adhering to data protection laws and industry standards ensures that
customer information remains secure. Implementing robust encryption, authentication
mechanisms, and stringent access controls safeguards against unauthorized access and data
breaches.

While the SmartCart project promises significant benefits, it's crucial to conduct a
comprehensive cost analysis. Assessing the expenses associated with developing, deploying,
and maintaining the system allows for an informed evaluation of its long-term viability and
return on investment. Careful financial planning will facilitate well-informed decisions for both
project developers and store owners.

Looking ahead, the SmartCart project has the potential to inspire further innovations in the
realm of retail technology. As the project gains traction and establishes its value proposition,
scalability becomes a consideration. Exploring opportunities to expand the SmartCart's reach
to different retail settings and markets opens avenues for growth and widespread adoption.

In conclusion, the SmartCart project exemplifies the power of technological innovation in

reshaping traditional practices. By seamlessly integrating object recognition, machine learning
algorithms, and a user-centric web application, the project not only enhances the shopping
experience but also empowers convenience store owners to optimize their operations and costs.

12
The journey to revolutionize the shopping journey is paved with challenges and opportunities,
ultimately redefining how customers interact with physical retail spaces.

1.2 Need Analysis

In today's retail landscape, the checkout process has often been a bottleneck, impacting both
customer satisfaction and store revenue. Recent studies highlight a prevailing sentiment among
92% of retail store owners who believe that long queues at billing points are significantly
detrimental to their business. This issue creates a two-fold problem, affecting both customers
and store owners, underscoring the need for innovative solutions.

From the customer's perspective, enduring lengthy queues during checkout can be a frustrating
experience. Time is a valuable commodity, and any time wasted in a queue feels unproductive
and arduous. Research findings indicate that 50% of shoppers rate the checkout experience as
"Very Important," further emphasizing the vital role of a seamless checkout in retaining
customers.

However, the implications of prolonged queues extend beyond mere annoyance. For store
owners, this situation presents an array of challenges that directly impact their operations and
profitability. These challenges include the following:

1. Space Constraints: Traditional checkout setups demand dedicated space for billing counters,
potentially occupying valuable real estate that could be utilized for more productive purposes,
such as expanding product displays or creating comfortable shopping areas.

2. Customer Dissuasion: Long queues can dissuade customers from shopping, resulting in
abandoned purchases and lost revenue opportunities. When confronted with the prospect of a
time-consuming checkout, some customers might opt to leave the store empty-handed.

3. Diminished Shopping Experience: A cumbersome checkout process diminishes the overall

quality of a customer's shopping journey. Even if customers enjoy the selection and ambiance
of a store, their experience can be tainted by a frustrating checkout.

13
Collectively, these issues culminate in a negative impact on sales, which, over time, can erode
customer loyalty. Establishing a reputation for consistently poor checkout experiences may
deter customers from returning to the store.

The SmartCart project presents a transformative solution to these challenges. By seamlessly

integrating cutting-edge technologies such as object recognition and real-time billing, the
SmartCart revolutionizes the checkout process. This innovation holds the potential to reduce
customer wait times significantly, as products are automatically billed the moment they are
placed in the cart. This not only enhances the efficiency of the checkout but also contributes to
an overall positive shopping experience.

1.3 Research Gaps

• Real-time Object Recognition in Shopping Carts: Despite the advancements in
object recognition technology, there is a lack of comprehensive research focused
specifically on real-time object recognition within shopping carts. Existing studies
often address object recognition in various contexts but not necessarily tailored to the
unique requirements and challenges of a shopping cart environment.

• Integrated Weight Detection and Billing Systems: While weight detection systems
are commonly used in industries like logistics and warehousing, their integration into
smart shopping carts for seamless billing and fraud prevention is an underexplored area.
Research on the accurate integration of weight detection systems with other cart
features and their practical deployment within retail settings is limited.

• Hybrid Recommendation Systems for Grocery Retail: Although hybrid

recommendation systems have gained attention in e-commerce, their specific
application within the grocery retail sector is relatively unexplored. Extending hybrid
approaches to account for the distinct characteristics of grocery shopping, such as
perishable items and dietary preferences, presents a research gap.

14
1.4 Problem Definition and Scope
The billing process in convenience stores is time-consuming, space-inefficient and expensive,
both for store owners and customers. Customers spend significant amounts of time waiting in
queues, while store owners have to allocate valuable store space and spend money on salaries
and equipment. The goal of this project is to improve the billing process in convenience stores
by reducing customer wait times, optimizing store space usage, and decreasing the costs
associated with the billing process.

The SmartCart project encompasses the creation of an intelligent shopping cart equipped with
object recognition technology for automatic billing and a companion mobile app for seamless
payments and personalized recommendations. The system aims to eliminate long queues,
enhance customer experience, and optimize store space by eliminating traditional billing
counters. It incorporates a weight-based system to prevent fraud. The project targets
operational efficiency by reducing reliance on manual scanning and potentially lowering
operational costs. Ensuring data privacy and security, along with regulatory compliance, is
essential. The SmartCart's goal is to revolutionize the shopping experience, increase customer
loyalty, and position the store as innovative. The project's scope includes technological
development, user education, pilot testing, and assessing long-term viability for potential
scalability.

1.5 Assumptions and Constraints

Sno Assumptions

1 Customers are able to use the self checkout smart grocery cart without assistance
and with ease.

2 Equipped with reliable and efficient object detection and weight measuring
technology.

3 Can accurately identify and match the scanned items with the items in the store's
inventory.

15
 4 Can provide customers with a clear and accurate receipt of their purchase.

5 Can be easily integrated with the store's existing inventory and point-of-sale
systems.

Table 1: List of Assumptions

S no Constraints

1 Object detection and weight measuring technology may not be 100% accurate.

2 May experience technical difficulties, such as connectivity issues or software

malfunctions.

3 May not be suitable for customers with disabilities or mobility issues.

4 For grocery stores only.

5 Ability to accurately count items may be affected by certain items, such as

produce that is sold by weight or items that are not clearly labelled.

Table 2: List of Constraints

1.6 Standards
Standards for Smart Cart with Object Recognition and Weight Detection

Accuracy and Reliability:

The smart cart system should have a high level of accuracy in both object recognition and
weight detection to ensure reliable performance.
Object recognition algorithms should be capable of identifying items with a high degree of
precision, reducing errors and false positives.

Real-Time Processing:
The system should operate in real-time, providing instant feedback to users for both object
recognition and weight detection.
Delays in processing and response times should be minimized to enhance user experience.

16
Interoperability:
The smart cart should be compatible with a wide range of products and items, ensuring that
various shapes, sizes, and materials can be accurately recognized.
The system should support industry-standard communication protocols to enable seamless
integration with existing retail infrastructure.

User-Friendly Interface:
The cart's user interface should be intuitive and easy to understand, providing clear feedback
on recognized items and their weights.
Visual displays and indicators should be easily visible to users during their shopping
experience.

Privacy and Security:

The system should prioritize customer privacy by anonymizing data collected during shopping.
Any data storage or transmission should be encrypted to prevent unauthorized access to
sensitive information.

Maintenance and Support:

The smart cart should be designed for ease of maintenance, with modular components that can
be replaced or upgraded as needed.
The manufacturer should provide comprehensive technical support and regular software
updates to address any issues and improve functionality.

Environmental Considerations:
The system should be energy-efficient to minimize its impact on the environment and ensure
prolonged battery life.
Materials used in the cart's construction should be environmentally friendly and recyclable.

Scalability:
The system should be designed with the flexibility to accommodate different store sizes and
layouts, from small convenience stores to large supermarkets.

17
Regulatory Compliance:
The smart cart should adhere to relevant industry and regulatory standards, ensuring safety and
compliance with local laws and regulations.

Cost-Effectiveness:
While incorporating advanced technologies, the smart cart should strive to be cost-effective,
providing value for both retailers and consumers.

Accessibility:
The user interface of the smart cart should be accessible to people with disabilities, ensuring a
more inclusive shopping experience for all.

Data Analytics and Insights:

The system should have the capability to collect and analyse shopping data, enabling retailers
to gain insights into consumer behaviour and preferences.

1.7 Approved Objectives

Integrating Object-Recognition System for Automated Billing:

Create an advanced object-recognition system that accurately identifies and bills products as
they are placed within the SmartCart, streamlining the checkout process for customers.

Incorporating Weight-Based Security System:

Implement a robust weight-based security system to mitigate shoplifting and fraudulent

activities, ensuring the integrity of the shopping experience and preventing unauthorized
removal of items from the cart.

Developing a User-Friendly Web Application:

Design and develop an intuitive web application catering to both customers and store
administrators, providing a seamless interface for effortless interactions, tracking, and
management of cart contents and transactions.

Enabling Personalized Product Recommendations:

18
Integrate an intelligent recommendation engine that leverages customer preferences and
shopping history, offering tailored product suggestions to enhance customer engagement and
satisfaction while potentially boosting sales.

1.8 Methodology
A smart shopping cart with a weight detection scale and a camera for object detection is a
complex system that requires careful planning and execution. The first step is to define the
requirements of the system. This includes determining the desired features, such as the weight
detection scale and camera for object detection. Once the requirements have been defined, the
hardware can be designed to meet these requirements. This involves selecting the appropriate
components and designing the system for ease of use and power efficiency.

The next step is to develop the software for the system. Real-time object detection is a computer
vision technique that involves identifying and localizing objects in an image or video in real-
time.

For this, we can explore the following systems and choose the most accurate out of the models.
Convolutional Neural Networks (CNNs) is a type of deep learning model that can be used for
real-time object recognition. They work by analyzing image features at different levels of
abstraction and using these features to classify objects. Single Shot Detector (SSD) is a popular
real-time object detection technique that can be used for object recognition as well. By
predicting the presence, location, and size of objects in real-time, it can recognize objects in a
video stream. Others include, You Only Look Once (YOLO) and Region-based Convolutional
Neural Networks (R-CNNs). YOLO is another real-time object detection algorithm based on a
single neural network. It works by dividing the input image into a set of grid cells and predicting
each cell's object class and bounding box. R-CNNs are a family of object detection algorithms
that work by generating region proposals in an image and then classifying the objects within
each proposal. Although they are not as fast as some of the other methods, they can achieve
high accuracy.

Another part is to program the weight detection scale to accurately measure the weight of items
added to the cart and program the camera to detect objects in the cart. The software should also
be designed to store data about the items added to the cart, such as the weight and item
description. The smart shopping cart should also be integrated with a backend system that can

19
store the data collected by the cart. This may include a cloud-based system or a local database.
The backend system should also be designed to interface with point-of-sale systems and
inventory management systems.

The final step is to deploy the smart shopping cart in a real-world environment, such as a
grocery store. This involves training store employees on how to use the system and monitoring
the system to ensure it functions correctly

1.9 Project Outcomes and Deliverables

Streamlined Shopping Experience:
Customers experience a faster and more convenient checkout process, reducing frustration and
wait times.

Enhanced Customer Engagement:

Personalized product recommendations increase customer engagement, leading to higher
transaction values and improved customer satisfaction.

Operational Efficiency:
Store operations become more efficient as the need for manual item scanning and traditional
billing counters is reduced.

Space Optimization:
Valuable store space is freed up by eliminating traditional billing counters, allowing for better
space allocation and visual merchandising.

Reduced Fraud and Shoplifting:

Implementation of the weight-based system effectively prevents shoplifting and unauthorized
item removal, ensuring the integrity of the shopping process.

Cost Savings:
Operational costs decrease due to reduced reliance on manual scanning and traditional
checkout counters, potentially leading to increased profitability.

20
Deliverables:

Object-Recognition System:
A functional object-recognition system capable of accurately identifying and billing products
placed in the SmartCart.

Weight-Based Security System:

A weight-based security mechanism integrated into the SmartCart, ensuring product integrity
and preventing fraud.

Companion Web Application:

A user-friendly web application accessible to both customers and store administrators, allowing
for seamless interaction with the SmartCart, transaction tracking, and cart management.

Product Recommendation Engine:

An intelligent recommendation engine integrated into the SmartCart and web application,
offering personalized product suggestions based on customer preferences.

1.10 Novelty of Work

• Integrated Weight Detection and Object Recognition: The novel aspect of our
project lies in the concurrent implementation of weight detection and real-time object
recognition within a smart shopping cart. This integration introduces a groundbreaking
approach to streamline the billing process and prevent fraudulent activities. By
accurately identifying items through object recognition and correlating them with
weight measurements, your system ensures both seamless checkout and inventory
management. This innovative combination of technologies enhances the cart's
functionality, demonstrating a pioneering solution for modern retail environments.

• Personalized Recommendations in Grocery Retail: The incorporation of

personalized product recommendations, based on hybrid recommendation systems,
tailored to the grocery retail sector represents a novel contribution. Few projects have
ventured into applying hybrid approaches specifically to the grocery domain,
showcasing the innovative nature of your work.

21
• Enhanced Shopping Cart Functionality: By incorporating features such as real-time
object recognition, weight detection, and mobile application integration, your smart
shopping cart goes beyond conventional carts. This enhanced functionality, coupled
with its potential to reduce wait times and offer tailored recommendations, sets your
project apart as a novel advancement in retail technology.

22
 Requirement Analysis

2.1 Literature Survey

Authors Title Year Description

[Link].

Smart shopping cart is

created with the help of radio
IoT-Based Smart
frequency identification
Mobeen Shopping Cart Using
1 2020 (RFID) sensors, an Arduino
Shahroz Radio Frequency
microcontroller, a Bluetooth
Identification
module, and a mobile
application.

Enhance the shopping

experience for customers by
Smart shopping cart reducing the need for human
for automated billing labor by automating the
Gangwal U.,
2 purpose using 2013 billing process. The
Roy S., Bapat J
wireless sensor customers can monitor the
networks details of their purchases and
view the current bill amount
on the screen.

Outlines how a Smart

Shopping Cart is developed
P.T.
using ZigBee networks. The
Sivagurunathan,
Smart Shopping design of the system is both
3 P. Seema, M. 2018
Trolley Using RFID cost-effective and
Shalini, R.
dependable, and it also
Sindhu
guarantees that deception is
detected.

Discusses collaborative
Xiaoyuan Su A Survey of
4 2009 filtering, which works on the
and Taghi M. Collaborative
assumption if two users have
Khoshgoftaar Filtering Techniques
purchased similar items in

23
 the past, they would prefer
similar items in the future.

Used the XGBoost Classifier

and the content-based
Product filtering approach on the Jeju
Zeinab
Recommendation online shopping mall dataset
5 Shahbazi and 2020
Based on Content- for product
Yung-Cheol
based Filtering Using recommendations and
Byun
XGBoost Classifier generated high-quality
recommendations.

Developed a state-of-the-art
Deep Learning deep learning
Recommendation recommendation model
Maxim Model for 2019 (DLRM) and provide its
6 Naumov et al Personalization and implementation in both
Recommendation PyTorch and Caffe2
Systems frameworks

H. -C. Chi, M. Deep learning networks such

Smart Self-Checkout
A. Sarwar, Y. - as Faster R-CNN, YOLOv2,
Carts Based on Deep
7 A. Daraghmi, 2020 and YOLOv2-Tiny are
Learning for
K. -W. Lin, T. - utilized to analyse the
Shopping Activity
U. Ik and Y. -L. content of each video frame.
Recognition
Li

A sensor will be placed at

the back of the ceramic tiles
that notifies the owner
R. B. Ramli, N.
The Development of through a message sent to
D. Afrina binti
8 Weight Detection 2021 the owner’s mobile. This
Nazri, O. I. Al-
System using IOT system is able to measure
Sanjary and N.
Flooring changes in pressure
Rozzani
according to the weight of an
object.

Discovered payment
9 C. Shuran and A New Public 2020 methods, putting forward a
Y. Xiaoling Transport Payment
system design scheme as the

24
 Method Based on fourth party payment
NFC and QR Code platform based on NFC
technology, QR code
technology, and combined
with IC card as well.

Tried to integrate two

Integration Model of payment gateways using
M. Hisyam and Multiple Payment core API integration to
10 2022
I. B. K. Gateways for Online provide a better experience
Manuaba Split Payment and more payment methods
Scenario to a user.

Table 3: Literature Survey Papers

2.1.1 Theory Associated with Problem Area

Throughout history, people have consistently developed technology to enhance efficiency,
particularly in tasks like shopping for essential items. To tackle the challenges of manual
billing using traditional methods like pen, paper, and barcode systems, a novel concept of
smart shopping carts utilizing modern technology was introduced. These innovative carts
incorporate advanced features and technologies to elevate the shopping experience for
customers.

2.1.2 Existing Systems and Solutions

In one instance, discussed in Paper [3], a smart shopping cart is created by integrating radio
frequency identification (RFID) sensors, an Arduino microcontroller, a Bluetooth module,
and a mobile application. This setup enables data transmission via Bluetooth from the
microcontroller to the linked mobile device, and subsequently to a server. Paper [4] focuses
on improving the shopping experience by automating the billing process, reducing the need
for manual labor. Customers can monitor their purchases and view the ongoing bill amount
on a display. Furthermore, Paper [5] delves into the real-time acquisition of information
about products within the shopping cart. Another approach outlined in Paper [6] utilizes
ZigBee networks to develop a cost-effective and reliable Smart Shopping Cart system.
This design ensures both buyer and seller appeal while utilizing ZigBee as a more
dependable alternative to the Xampp server.

25
2.1.3 Research Findings for Existing Literature

In conclusion, smart shopping carts are increasingly adopted by retailers to enhance

customer shopping experiences. However, challenges such as cost and privacy concerns
must be considered. Despite these challenges, the use of smart shopping carts is projected
to grow as retailers seek to enhance efficiency, boost sales, and offer a more convenient
and personalized shopping experience for customers.

2.1.4 Problem Identified

Product recommendation systems aid shoppers in discovering new items, minimizing

search time, and improving the overall shopping process. A prevalent method for product
recommendation is collaborative filtering.

2.1.5 Survey of Tools and Technologies Used

In summation, collaborative filtering, content-based filtering, and hybrid recommendation

systems have all demonstrated effectiveness in product recommendation within the
grocery industry. Advancements in deep learning are expected to further enhance accuracy
and effectiveness in these systems. Additionally, weight detection systems in smart carts
offer valuable applications, particularly in settings like grocery stores or warehouses.

For instance, one common approach for weight detection involves using load cells with
amplifiers like HX711. Another method explores utilizing cameras and machine learning
algorithms for weight estimation based on visual features. Object recognition and deep
learning have gained attention in smart shopping cart development, facilitating accurate
identification of items placed in the cart.

A promising alternative to traditional barcode scanning is the use of RFID readers. These
readers scan products as customers place them in their carts, with the information
displayed on a cart screen and sent to a database. This method, although requiring
technological updates, has the potential to expedite scanning and enhance the shopping
experience.

Furthermore, the integration of NRF24L01 wireless communication and LCD displays has
been proposed to facilitate seamless communication between the shopping basket and the
cashier, improving efficiency in all store areas.

26
2.2 Software Requirement Specification

2.2.1 Introduction

[Link] Purpose
The billing process in convenience stores is time-consuming, space-inefficient and
expensive, both for store owners and customers. Customers spend significant amounts
of time waiting in queues, while store owners must allocate valuable store space and
spend money on salaries and equipment. The goal of this project is to improve the
billing process in convenience stores by reducing customer wait times, optimizing store
space usage, and decreasing the costs associated with the billing process.

[Link] Intended Audience

The SmartCart project's intended audience encompasses a diverse range of individuals

and groups. Consumers seeking a seamless shopping experience are a primary focus,
benefiting from the project's streamlined checkout process and convenience.
Convenience store owners stand to gain operational efficiency and reduced costs,
making them another crucial audience. Technology enthusiasts intrigued by innovative
integration of object recognition, machine learning, and app development will find the
project captivating. Retail professionals aiming to enhance customer experiences and
investors seeking promising ventures in the retail tech sector are also important
audiences. Additionally, tech communities, regulatory authorities, and media outlets
are interested in the project's advancements, impact on the retail landscape, and
potential for transformative change.

[Link] Project Scope

The SmartCart project's scope encompasses the development of cutting-edge object

recognition technology for automatic billing, a companion web app for cashless
payments and recommendations, and a weight-based system to prevent fraud. It aims
to streamline the shopping experience by reducing checkout time, optimizing store
space, and cutting costs. Technical components include machine learning algorithms,
user-friendly app development, and data privacy measures. Challenges like accurate
recognition and seamless functionality are acknowledged. The project's broader impact

27
 spans technological innovation, operational efficiency, and transforming the retail
landscape.

2.2.2 Overall Description

[Link] Product Perspective

From a product perspective, SmartCart introduces an innovative paradigm shift to the
shopping experience, featuring a smart shopping cart as its core physical component.
This cart leverages cutting-edge object recognition technology to seamlessly identify
and bill items as they are placed within it, eliminating the need for conventional barcode
scanning and expediting the checkout process significantly.

Complementing the smart cart is a web application that enhances user engagement and
convenience. This web app enables cashless payments, delivers personalized product
recommendations grounded in shopping history, guides patrons to specific items within
the store layout, and manages dynamic shopping lists. The web app acts as an accessible
and user-centric interface, facilitating smooth interaction between customers and the
cart.

In order to thwart fraudulent activities, a robust weight-based system is integrated into

the solution. This system cross-references the combined weight of cart items against
the anticipated weight calculated from scanned products, ensuring precision in billing
accuracy and security enhancement.

From a technical standpoint, the project involves the development of sophisticated

machine learning algorithms to power the recommendation engine and the creation of
an intuitive and secure web app interface. Prioritizing data privacy and security is
pivotal in cultivating user confidence. While challenges tied to precise object
recognition and seamless performance are acknowledged, the SmartCart product is
poised to redefine retail paradigms by harmonizing cutting-edge technology,
operational efficiency, and customer satisfaction.

[Link] Product Features

SmartCart boasts an array of innovative and user-centric features aimed at enhancing

the shopping experience while revolutionizing convenience stores' operational
efficiency:

28
• Automatic Billing Through Object Recognition: The core feature allows the smart
shopping cart to automatically identify items as they're placed inside, eliminating the
need for manual scanning. This expedites the checkout process and enhances customer
convenience.
• Companion Web Application: The web app provides a user-friendly interface that
enables cashless payments, personalized product recommendations based on shopping
history, guided navigation to store items, and real-time management of shopping lists.
• Weight-Based Fraud Prevention: A weight-based system enhances security by
comparing the cart's total weight against the theoretical weight of scanned items. This
ensures accurate billing and minimizes the potential for shoplifting.
• Efficient Store Navigation: The app guides customers to specific items within the
store, enhancing their shopping experience by reducing search time and increasing
product visibility.
• Personalized Recommendations: Through machine learning algorithms, the system
offers tailored product recommendations, helping customers discover new items
aligned with their preferences and past purchases.
• Real-Time Inventory Tracking: SmartCart keeps track of inventory in real time,
providing store owners with valuable insights into stock levels and popular products,
aiding in inventory management.
• Cashless Payments: Customers can make secure and convenient cashless payments
through the app, reducing reliance on physical payment methods and expediting the
checkout process.
• Convenient Shopping Lists: The app's dynamic shopping list feature allows users to
maintain and update their lists as they shop, ensuring they never forget essential items.
• Operational Cost Reduction: By automating the billing process and reducing the need
for manual scanning, SmartCart helps convenience store owners save on labor costs
and operational resources.
• Space Optimization: With the elimination of traditional billing counters, stores can
allocate space more efficiently, maximizing the store's layout for enhanced customer
flow.
• Enhanced Customer Engagement: The personalized recommendations and guided
shopping experience foster greater engagement, encouraging customers to explore
more products and make additional purchases.

29
• User Privacy and Security: Strong data protection measures are embedded within the
system to ensure customer privacy and secure transactions.

By combining these features, SmartCart not only modernizes the shopping journey but
also addresses the needs of both customers and convenience store owners, resulting in
a win-win solution that redefines the retail landscape.

2.2.3 External Interface Requirements

[Link] User Interfaces

The user is required to have a mobile phone with a camera to scan the QR code on the
cart and Internet connectivity to access the web application.

[Link] Hardware Interfaces

The SmartCart project necessitates a core set of hardware components for its successful
implementation. This includes weight sensors (load cells) to enable accurate billing and
prevent fraud, a Raspberry Pi or Arduino as the central processing unit, a battery for
wireless mobility, a camera for object recognition, and a physical cart to house and
integrate these elements. An alarm system with a buzzer enhances security, while
connecting wires and potential printed circuit boards (PCBs) facilitate seamless
communication between components. These hardware essentials collectively realize
SmartCart's objectives of automated billing, fraud prevention, and enhanced customer
experience.

[Link] Software Interface

The successful implementation of the SmartCart project relies on a range of external

software components. These include specialized object recognition software to identify
items in the cart, a machine learning framework for personalized recommendations and
fraud prevention, and web development tools for creating a user-friendly web
application interface. Database management systems are essential for storing product
data and transaction history securely. Security software ensures data protection, while
communication libraries enable real-time interactions between the cart and the app. The
Raspberry Pi requires an operating system, and integration with payment gateways
facilitates cashless transactions. Image processing libraries handle captured images,

30
 while version control, collaboration tools, and deployment services streamline
development and launch. Lastly, data analytics tools aid in understanding customer
behavior. These components collectively form the foundation of SmartCart's
functionality, security, and user experience.

2.2.4 Other Non-functional Requirements

[Link] Performance Requirements

The SmartCart project's performance requirements encompass a range of critical aspects

to ensure its efficiency and effectiveness. These include achieving high object
recognition accuracy, maintaining real-time responsiveness during shopping and
checkout, delivering swift and personalized product recommendations, and ensuring
seamless navigation within the store. Security and fraud prevention mechanisms must
work seamlessly, while data privacy standards need to be rigorously upheld. The system's
reliability, availability, scalability, and user-friendly interface are paramount. Battery life
should support uninterrupted operation, and integration with existing store systems
should occur seamlessly. Compatibility with various devices and browsers, effective
error handling, and a framework for regular maintenance and updates are also crucial. By
adhering to these performance benchmarks, SmartCart can provide a reliable, user-
centered shopping solution that benefits both customers and store owners.

[Link] Safety Requirements

Safety is paramount in the SmartCart project, with specific safety requirements

embedded throughout the system. The object recognition technology must accurately
identify items to prevent billing errors that could impact customers. The weight-based
system and fraud prevention measures ensure a secure shopping environment,
minimizing the potential for theft or unauthorized transactions. Data privacy safeguards
protect customer information, and error handling mechanisms swiftly address any
anomalies. The physical cart design prioritizes stability and maneuverability, reducing
the risk of accidents. The system also undergoes rigorous testing to guarantee its safe and
reliable operation within the store environment. By adhering to these safety requirements,
SmartCart ensures a secure and worry-free shopping experience for all users.

31
 [Link] Security Requirements

Security is a fundamental consideration in the SmartCart project, with stringent security

requirements integrated into its design. The system employs robust authentication and
encryption protocols to safeguard customer data and payment information during
cashless transactions. Access controls are implemented to prevent unauthorized access
to sensitive components. The object recognition and weight-based systems work in
tandem to prevent fraudulent activities and ensure accurate billing. Regular security
audits and updates are conducted to identify and mitigate vulnerabilities. The entire
solution adheres to industry best practices and standards to provide a secure shopping
environment for customers and maintain the integrity of the system.

2.3 Cost Analysis

Figure 1: Cost Analysis

2.4 Risk Analysis

Risk Analysis for the SmartCart Project:

• Technical Challenges: The integration of object recognition and machine learning

algorithms poses the risk of technical complexities leading to project delays and budget
overruns. Mitigation involves thorough prototyping, expert consultation, and allocating
sufficient time for R&D.
• Data Privacy and Security: Inadequate security measures may result in data breaches,
harming customer trust and leading to legal consequences. Risk mitigation includes
robust encryption, regular security assessments, and compliance with data protection
regulations.
• False Positives/Negatives: Inaccuracies in object recognition could cause erroneous
billing or missed items, frustrating customers and causing financial discrepancies. Risk
management involves continuous testing, refining algorithms, and implementing fail-
safe mechanisms.

32
 • Integration Complexity: Challenges in integrating hardware components (weight
sensors, Raspberry Pi, camera) with the web application might lead to compatibility
issues or malfunctions. Thorough testing and validation of components and interfaces
are critical for risk mitigation.
• User Adoption: Customer resistance or usability issues may hinder adoption. Risk
mitigation requires user-friendly interfaces, comprehensive user training, and support
mechanisms to address customer concerns.
• Operational Interruptions: Technical failures during peak hours could disrupt store
operations and lead to dissatisfaction. Mitigation involves redundancy planning, regular
system maintenance, and quick response to technical issues.
• Regulatory Compliance: Failure to adhere to data privacy laws and regulations could
result in legal penalties. Compliance audits, legal expertise, and proactive data
protection measures are necessary to mitigate this risk.
• Maintenance and Updates: Neglecting timely software updates and maintenance may
expose the system to security vulnerabilities or performance issues. Risk mitigation
includes a well-defined maintenance plan and regular updates.
• Unforeseen Costs: Unexpected expenses could strain the project budget. Contingency
funds, thorough cost estimation, and regular financial reviews are essential for risk
mitigation.
• Competition and Market Dynamics: Shifts in market trends or competitive technologies
might impact adoption rates. Continuous market analysis, agility in adapting to
changes, and customer-centric improvements can help mitigate this risk.

To navigate these risks effectively, a comprehensive risk management strategy should be

implemented. Regular risk assessment, contingency planning, and adaptability to unforeseen
challenges will contribute to the project's success and resilience.

33
 Methodology Adopted

3.1 Investigative Techniques

Introduction:
The integration of technology into everyday life has transformed various industries, including
retail. The smart shopping cart is a cutting-edge innovation aimed at enhancing the shopping
experience by leveraging object recognition and weight sensor technologies. This document
aims to justify the selection of these investigative techniques and their integration into the smart
shopping cart project.

Object Recognition:
Object recognition technology has rapidly evolved with advancements in artificial intelligence
and computer vision. By employing object recognition, the smart shopping cart can accurately
identify and classify items placed within it. This is crucial for an efficient and seamless
shopping experience. Traditional barcode scanning or manual entry of items can be time-
consuming and error-prone. Object recognition eliminates these drawbacks by automatically
identifying items based on their visual attributes, such as shape, color, and packaging.

Weight Sensor:
Weight sensors are another crucial aspect of the smart shopping cart project. These sensors are
integrated into the cart's platform and enable real-time monitoring of the weight changes as
items are added or removed. The weight sensor technology complements object recognition by
providing an additional layer of verification and security.

Integration of Object Recognition and Weight Sensor:

The integration of object recognition and weight sensor technologies synergistically enhances
the smart shopping cart's capabilities. Object recognition technology efficiently identifies
items, while weight sensors validate their presence and quantity. This comprehensive approach
reduces the likelihood of errors and enhances security, making it a valuable asset for both
customers and retailers.

34
Item Verification: Weight sensors play a vital role in verifying the items placed in the cart.
Each item has a distinct weight, allowing the system to cross-reference the recognized items
with their expected weights. Any discrepancies between the recognized items and their weights
can trigger alerts, helping to prevent theft or accidental item placement.

Enhanced Checkout Process: At the checkout counter, the weight sensors contribute to a
smooth and expedited process. The system can automatically tally the total purchase value
based on the combined weights of the recognized items. This minimizes the time customers
spend at the checkout, leading to shorter queues and an improved shopping experience.

Accuracy and Efficiency: Object recognition technology has demonstrated impressive

accuracy rates in identifying a wide range of products. This accuracy contributes to an efficient
shopping process, as customers can simply place items in the cart without the need for manual
input. Additionally, it reduces the risk of errors associated with manual item scanning, resulting
in improved customer satisfaction.

Adaptability: Object recognition can be trained to recognize new products over time. This
adaptability ensures that the smart shopping cart remains relevant and up-to-date with changing
product offerings. It can also assist in providing personalized shopping recommendations based
on the recognized items, enhancing customer engagement.

Conclusion:
In conclusion, the selected investigative technique of combining object recognition and weight
sensor technologies for the smart shopping cart project is well-justified. Object recognition
ensures accurate item identification, while weight sensors provide verification and security.
The integration of these techniques aligns with the goals of improving efficiency, accuracy,
and customer experience in the retail environment. As technology continues to advance, the
smart shopping cart stands as a prime example of how innovation can transform traditional
processes into seamless, automated experiences.

The investigative techniques chosen for this project are data analysis, object recognition using
YOLOv8 model, and weight sensors. Each technique has been selected for its unique
contribution to the project's success.

35
Data Analysis: The project leverages data analysis techniques to manage the items within the
smart shopping cart. By using a MongoDB database, the system effectively stores and retrieves
data about the items added to the cart. This approach ensures accurate tracking of items and
their quantities, allowing for real-time updates to the cart's content.

Object Recognition (YOLOv8 Model): The YOLOv8 model for object recognition plays a
pivotal role in identifying items placed within the cart. This technique enables the system to
classify items and distinguish between products within the cart and those outside. The model's
efficiency and accuracy contribute to a seamless shopping experience for users, as it aids in the
automatic virtual noting of items and prevents any discrepancies.

Weight Sensors: The integration of weight sensors serves multiple purposes. Firstly, they
enhance security by detecting any attempts to sneak items into the cart without proper scanning.
Secondly, weight sensors provide an additional layer of accuracy in tracking the contents of
the cart. The combination of weight sensor data and object recognition ensures the virtual cart's
accuracy and reliability.

3.2 Proposed Solution

The proposed solution for the smart shopping cart project is aimed at revolutionizing the
shopping experience by incorporating advanced technologies:

Problem Analysis: The challenge lies in creating an efficient and reliable system that automates
the process of tracking items within a shopping cart while ensuring security against theft.

Proposed Solution: The solution involves the integration of object recognition using the
YOLOv8 model, weight sensors, and a companion application. The YOLOv8 model identifies
items within the cart, and weight sensors add a security layer. The MongoDB database keeps
track of item quantities. The companion application enables users to connect to the cart's
database using QR codes.

36
Benefits of the Solution: This solution offers improved accuracy in tracking items, prevents
theft, enhances the shopping experience, and reduces manual effort for customers and store
staff.

Implementation Strategy: The implementation will occur in stages: refining the object
recognition model, integrating weight sensors with the system, developing the companion
application, and ensuring seamless communication between the cart and the companion app.

Timeline and Milestones:

Refining Object Recognition: 2 months

Integrating Weight Sensors: 1.5 months
Developing Companion App: 2.5 months

3.3 Work Breakdown Structure (WBS) and Discussion on Workable

Modules/Products

Research and Development:

The initial phase of the project involves exploring the capabilities of the YOLOv8 model and
understanding various techniques for integrating weight sensors.

Exploring YOLOv8 Model:

This sub-task entails researching and understanding the YOLOv8 model's capabilities,
architecture, and object recognition accuracy. This exploration helps lay the foundation for
implementing the object recognition module effectively.

Studying Weight Sensor Integration Techniques:

In this sub-task, various techniques for integrating weight sensors into the smart shopping cart
are researched. The focus is on understanding how weight data can be synchronized with object
recognition data to enhance security and accuracy.

37
Object Recognition and Database Integration:
This phase involves developing the object recognition module using the YOLOv8 model and
integrating it with the MongoDB database.

Developing YOLOv8 Model for Object Recognition:

In this sub-task, the YOLOv8 model is implemented to identify and classify items within the
cart. The model is trained using relevant datasets to accurately detect various products.

Integrating YOLOv8 with the MongoDB Database:

Here, the recognized items are linked to the MongoDB database. This integration ensures that
real-time updates of cart contents are reflected in the database, creating a seamless shopping
experience.

Weight Sensor Integration:

This phase focuses on incorporating weight sensors into the smart shopping cart and ensuring
their integration with the object recognition data.

Procuring and Calibrating Weight Sensors:

The task involves procuring suitable weight sensors and calibrating them to ensure accurate
weight measurements. Proper calibration is essential for reliable security and verification.

Integrating Weight Sensor Data with Object Recognition Data:

This sub-task involves developing algorithms to combine weight sensor data with object
recognition data. The system will verify recognized items based on their expected weights,
enhancing both security and accuracy.

Companion Application Development:

The development of a user-friendly companion application that interacts with the smart
shopping cart is the focus of this phase.

Designing User Interface:

This sub-task involves creating an intuitive and visually appealing user interface for the
companion application. The design ensures ease of use and a seamless experience for
customers.
38
Implementing QR Code Scanning and Database Communication:
Here, the functionality to scan QR codes on the carts and establish communication with the
cart's database is developed. This feature enables customers to access real-time data and review
their shopping lists conveniently.

Discussion on Workable Modules/Products:

Object Recognition Module:

The object recognition module powered by the YOLOv8 model serves as the foundation of the
smart shopping cart's functionality. By identifying items in real-time, this module ensures that
customers' selections are accurately recorded. It provides a seamless and efficient shopping
experience by eliminating the need for manual input and barcode scanning. The integration
with the MongoDB database ensures that recognized items are securely stored and can be easily
accessed for tracking and checkout.

Weight Sensor Module:

The weight sensor module adds a layer of security and verification to the smart shopping cart.
By procuring and calibrating weight sensors, the system can accurately determine changes in
cart weight. This integration complements the object recognition module, enhancing security
by preventing unauthorized removal of items and promoting accuracy by cross-referencing
recognized items with their expected weights. The combined efforts of both modules ensure a
trustworthy shopping experience.

Companion Application:
The companion application expands the smart shopping cart's capabilities beyond the physical
cart. With a user-friendly interface and QR code scanning functionality, the app establishes a
direct link to the cart's database. Customers can conveniently review their selected items, track
their shopping progress, and receive real-time updates. This engagement promotes a
personalized shopping journey and streamlines the overall experience.

In conclusion, the Work Breakdown Structure emphasizes the different stages of the project,
from research and development to the creation of workable modules. The Object Recognition
Module, Weight Sensor Module, and Companion Application collectively contribute to an
39
innovative smart shopping cart solution that enhances security, accuracy, and customer
engagement. The integration of YOLOv8, weight sensors, and a companion app showcases the
project's holistic approach to modernizing the shopping experience.

3.4 Tools and Technology

Tools and Technologies Used and Their Justification:

1. YOLOv8 Model for Object Recognition:

Tool/Technology: YOLOv8 (You Only Look Once version 8) is a deep learning model known
for its real-time object detection capabilities.
Justification: YOLOv8 is chosen for its ability to rapidly and accurately identify objects in real-
time video streams. Its single-pass approach enables efficient processing, aligning with the
need for quick and seamless object recognition within the smart shopping cart. This technology
significantly enhances the shopping experience by eliminating the need for manual barcode
scanning or input.

2. Weight Sensors for Security and Verification:

Tool/Technology: Weight sensors are hardware devices used to measure the weight of objects
placed on them.
Justification: Weight sensors are essential for the security and accuracy of the smart shopping
cart system. They prevent unauthorized removal of items by detecting weight changes and
verify recognized items by cross-referencing their expected weights. The choice of weight
sensors ensures the system's reliability and helps maintain trust between customers and the
technology.

3. MongoDB Database for Data Storage:

Tool/Technology: MongoDB is a NoSQL database that provides flexibility and scalability for
handling unstructured or semi-structured data.
Justification: MongoDB's document-based structure is well-suited for storing recognized items
and virtual notes within the smart shopping cart system. It offers real-time data synchronization
and easy data retrieval, crucial for maintaining accurate and up-to-date information about cart

40
contents. The flexibility of MongoDB accommodates changes in data structure, making it
suitable for a dynamic environment like a shopping cart.

4. Mobile Application Development Platform:

Tool/Technology: A mobile application development platform, such as React Native or Flutter,
is used to create the companion application.
Justification: These platforms allow for the development of cross-platform applications,
ensuring compatibility with both Android and iOS devices. This cross-platform approach
reduces development time and efforts. The mobile application provides customers with an
intuitive interface for interacting with the smart shopping cart, enhancing user engagement and
convenience.

5. QR Code Scanning Technology:

Tool/Technology: QR code scanning technology, integrated into the companion application,
enables users to connect with the cart's database.
Justification: QR codes provide a quick and convenient method for customers to establish a
connection with the smart shopping cart. Scanning a QR code eliminates the need for manual
entry and simplifies the process of accessing real-time cart data. This technology enhances the
user experience and encourages interaction with the companion application.

6. Python Programming Language:

Tool/Technology: Python is used for developing the object recognition module, integrating
YOLOv8, and performing data processing tasks.
Justification: Python is widely recognized for its ease of use, extensive libraries, and robust
support for machine learning and computer vision tasks. Its versatility allows for rapid
development and experimentation, which is crucial for implementing the object recognition
module effectively.

7. Calibration Tools for Weight Sensors:

Tool/Technology: Calibration tools are utilized to ensure the accuracy and reliability of the
weight sensors.
Justification: Calibration is essential to guarantee that weight measurements are accurate and
consistent. Calibration tools enable fine-tuning and adjustment of weight sensors, ensuring they
provide reliable data that aligns with the actual weight of items placed on the cart.
41
8. User Interface (UI)/User Experience (UX) Design Tools:
Tool/Technology: UI/UX design tools like Adobe XD, Sketch, or Figma are used to create
intuitive and visually appealing interfaces for both the cart's display and the companion
application.
Justification: These tools help designers visualize and prototype user interfaces, ensuring that
the final products are user-friendly, aesthetically pleasing, and aligned with modern design
principles.

In conclusion, the selection of tools and technologies is a crucial aspect of the smart shopping
cart project's success. Each tool or technology is chosen based on its suitability for the specific
tasks it will perform, ensuring efficient object recognition, accurate weight verification,
seamless database integration, user-friendly interaction, and reliable data storage. The
combination of these tools and technologies forms a cohesive ecosystem that enhances the
overall shopping experience and security for both customers and retailers.

42
 Design Specifications

4.1 System Architecture

Figure 2: System Architecture Diagram

4.2 Design Level Diagrams

Figure 3: Design Level Diagram

43
4.3 User Interface Diagrams

Figure 4: User Interface Diagram

4.4 Snapshots of Working Prototype

44
 Figure 5: Snapshot of the prototype detection the object

Figure 6: Snapshot of the prototype adding the object to the bill by passing the line

Figure 7: Snapshot of the Database after the object is added to the bill

45
 Conclusions and Future Scope

5.1 Work Accomplished

• Developed a machine learning model using YOLOv8 for object recognition.

• Implemented a mechanism to detect when an object crosses the middle of the screen,
triggering recognition and tracking of items.

5.2 Conclusion

The project combines advanced technologies like object recognition, machine learning, and
weight-based security to create an innovative and seamless shopping experience. By
automating billing, providing personalized recommendations, and enhancing convenience, you
are transforming traditional shopping into a tech-driven, efficient process.

5.3 Environmental (/ Economic/ Social) Benefits

Environmental Benefits:

• Most receipts are printed on thermal paper which has a chemical lining, making it
unrecyclable. [[Link] Our Companion web
app will digitalize receipts, reducing the amount of unrecyclable paper being used.
• Traditional checkout counters often require energy-intensive equipment. By
eliminating the need for these counters and streamlining the checkout process, your
project can lead to energy savings.

Economic Benefits:

• With the SmartCart, the amount of active personnel (staff responsible for billing)
required on hand will reduce, reducing working costs for store owners.
• The amount of time people spends in queues, waiting for their cart to be billed will
reduce, improving the customer’s experience. An improved customer experience will
help store owners repeat business.
• The product recommendation system on the webapp will bring more products to the
customer’s attention, directly increasing sales.

46
Social Benefits:

• Enhanced Shopping Experience: Your project aims to provide customers with a

seamless and efficient shopping experience. This can enhance customer satisfaction and
make shopping more enjoyable.
• Reduced Checkout Wait Times: By eliminating the need to wait in long checkout lines,
customers can save time during their shopping trips. This can be particularly helpful for
busy individuals and families.
• Personalized Recommendations: Providing personalized product recommendations can
help customers discover new products that align with their preferences, leading to a
more tailored and enjoyable shopping experience.

5.4 Future Work Plan

The future work plan mainly has the following phases

• Creating the companion webapp and hosting it. (Front end interface, database schema,
back end)
• Building the prototype of the smart cart itself. (Implementing the camera, load cells and
onboard computer)
• Integrating data received from sensors on the cart and the machine/deep learning
algorithms with the back end of the web app.

47
 APPENDIX A: References

[1]:[Link]
retailerrevenue/

[2]: [Link]

[3] Mobeen Shahroz, Muhammad Faheem Mushtaq, Maqsood Ahmad1, Saleem Ullah, Arif

Mehmood, And Gyu Sang Choi “IoT-Based Smart Shopping Cart Using Radio Frequency

Identification”, 2020

[4] Gangwal, U., Roy, S., Bapat, J.: Smart shopping cart for automated billing purpose using

wireless sensor networks. IEEE (2013)

[5] Raju Kumar, K. Gopalakrishna, K. Ramesha, “Intelligent Shopping Cart,” International

Journal

of Engineering Science and Innovative Technology (IJESIT) Volume 2, Issue 4, July 2013.

[6] P.T. Sivagurunathan, P. Seema, M. Shalini, R. Sindhu Smart Shopping Trolley Using RFID

International Journal of Pure and Applied Mathematics (2018).

[7] Xiaoyuan Su and Taghi M. Khoshgoftaar, “A Survey of Collaborative Filtering

Techniques”,

2009

[8] S. Gong, H. Ye and Y. Dai, "Combining Singular Value Decomposition and Item-based

Recommender in Collaborative Filtering," 2009

[9] Robin van Meteren and Maarten van Someren “Using Content-Based Filtering for

Recommendation”, 2016

[10] Zeinab Shahbazi and Yung-Cheol Byun “Product Recommendation Based on Content-
based

48
Filtering Using XGBoost Classifier”, 2020

[11] Duen-Ren Liu, Chin-Hui Lai and Wang-Jung Lee, “A hybrid of sequential rules and

collaborative filtering for product recommendation”, 2009

[12] Maxim Naumov et al, “Deep Learning Recommendation Model for Personalization and

Recommendation Systems”, 2019

[13] H. -C. Chi, M. A. Sarwar, Y. -A. Daraghmi, K. -W. Lin, T. -U. Ik and Y. -L. Li, "Smart
SelfCheckout Carts Based on Deep Learning for Shopping Activity Recognition," 2020 21st
Asia-Pacific Network Operations and Management Symposium (APNOMS), Daegu, Korea
(South),

2020, pp. 185-190, doi: 10.23919/APNOMS50412.2020.9237053.

49
APPENDIX B: Plagiarism Report

50