PROJECT PHASE-I

REPORT ON

“Suspicious Person Detection using Infrared Camera”

A Project Phase I report submitted in partial fulfillment of the requirement for
the award of Degree of

Bachelor of Technology
(Seventh Semester)
In
COMPUTER SCIENCE AND ENGINEERING
Session 2024-2025

Prescribed By
DBATU University, Lonere
Guided By Submitted By
Prof.Manisha More 1. DIPTI BANGDE(CSEA707)
2. AFIYA MOHAMMAD(CSEA702)
3. TANZILA SHEIKH(CSEB746)
4. ADITI CHARLAWAR(CSEA721)
5. HRUTUJA TIPLE(CSEA754)
 DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING
RAJIV GANDHI COLLEGE OF ENGINEERING
RESEARCH & TECHNOLOGY,
CHANDRAPUR
Session 2024-2025

CERTIFICATE

This is to clarify that, Ms. DIPTI BANGDE(CSEA707), Ms. AFIYA MOHAMMAD(CSEA702), Ms.
TANZILA SHEIKH(CSEB746) ,Ms. ADITI CHARLAWAR(CSEA721), Ms. HRUTUJA
TIPLE(CSEA754) studying in seventh semester of computer science and Engineering
Department.

In this session 2024-2025.

Have completed the project phase I

“Suspicious Person Detection using Infrared Camera”

Satisfactorily during the academic session 2024-2025 from Rajiv Gandhi College of
Engineering Research & Technology, Chandrapur.

PROF.MANISHA MORE Dr. MANISHA PISE

Project Guide Project in charge

Dr. Nitin Janwe

H.O.D.

Computer Science and Engineering

Chandrapur
 Institute Vision

To be on forefront to impart quality education to address societal and

industrial needs and imbibe career skills through perseverance and
practice.

Institute Mission

M1. To adapt innovative student centric learning methods based on

understanding and practice.

M2.To enhance professional and entrepreneurial skills.

M3.To motivate students to meet dynamic needs of the society with novelty
and creativity.

M4.To promote research and continuing education to keep the country ahead.

M5.To promote the mindset to acquire local solutions to local

problems (LS2LP).

Department Vision
To be a centre of excellence in Computer Science & Engineering by imparting
knowledge, professional skills and human values.

Department Mission
M1. To create encouraging learning environment by adapting innovative
student centric learning methods promoting quality education and
research.
M2. To make students competent professionals and entrepreneurs by
imparting career skills and ethics.
M3. To impart quality industry oriented education through industrial
internships, industrial projects and partnering with industries to
make students corporate ready.
Program Educational Objectives

PEO1: Impart quality industry oriented education, strong fundamentals

and problem solving approach through student centric learning
methods for product development and handle real time problems.
PEO2: Possess good communication skills and ethics in line with
corporate environments to serve the industry & society.
PEO3: Create Competent Professionals & entrepreneurs by imparting
Career skills, leadership qualities and research

Program Specific Outcomes:

PSO1: Foundation of Computer System: Ability to understand the

principles and working of computer systems. Students can assess
the hardware and software aspects of computer systems.

PSO2: Foundations of Software development: Ability to understand the

structure and development methodologies of software systems.
Possess professional skills and knowledge of software design
process. Familiarity and practical competence with a range of
programming languages and platforms like Cloud Computing,
Web based and Mobile applications, Image and Video Processing,
Artificial Intelligence & Machine Learning.

PSO3: Foundation of mathematical concepts: Ability to apply

mathematical methodologies to solve computational tasks, model
real world problems using appropriate data structures and suitable
algorithms.
 Rajiv Gandhi College of Engineering Research & Technology, Chandrapur
Department of Computer Science & Engineering

Project Title:- SENTIMENT ANALYSIS ON ONLINE PRODUCT REVIEW

Project PO/ PSO Attainment
PO/PSO Attainme Description
nt
(level1=lo
w, 2=
moderate,
3=High,
=not applicable)
In this project work, engineering knowledge is
PO1 3 applied at highest level.
In this project work, Problem analysis is
PO2 3 done at highest level.
In this project work, Design/Development of
PO3 3 Solutions is applied at highest level.
In this project work, conduction of
PO4 3 investigations of complex problem is done at
highest level.
In this project work, modern tool usuage is
PO5 3 applied at highest level.
In this project work, the engineer and society
PO6 3 concept is applied at highest level.
In this project work, the engineer and
PO7 society concept is not applicable.
In this project work, Ethics is not
PO8 applicable.
In this project work, Individual and
PO9 3 teamwork exists at highest level.
In this project work, Communication exists at
PO10 3 highest level.
In this project work, Project management
PO11 2
and finance exists at moderate level.
In this project work, Life Long Learning
PO12 3 exists at highest level.
In this project work, Foundation of
PSO1 3 computer System exists at highest level.
In this project work, Foundation of Software
PSO2 3 Development exists at highest
level.
In this project work, Foundation of mathematical
PSO3 3 concepts exists at highest level.

Signature of Project Guide

INDEX:
Chapters Index Page no.

1 Abstract 1

2 Introduction 2

3 Literature Review 3-4

4 Problem Statement 5

5 System Requirements 6

6 Technologies Used 7-8

7 Architecture of Project 9-10

8 Modules to be Developed 11-14

9 Code for Authentication Module 15-20

10 Output of Authentication Module 21

11 Code for UI Module 22-25

12 Output of UI Module 26

13 Advantages and Disadvantages 27-28

14 Conclusion 29

15 Bibliography 30
 1. ABSTRACT

Sentiment analysis on online product reviews has become a pivotal tool for businesses
seeking to understand customer opinions and improve their offerings. By leveraging natural
language processing (NLP) techniques, this approach classifies reviews as positive, negative, or
neutral, enabling organizations to derive actionable insights from unstructured text data. The
process involves steps such as data cleaning, feature extraction, and sentiment classification
using methods ranging from traditional machine learning algorithms like Support Vector
Machines (SVM) to advanced deep learning models such as BERT. These technologies allow
businesses to analyze customer feedback efficiently, uncovering valuable patterns and trends.

Beyond classification, sentiment analysis provides strategic advantages by highlighting

product strengths and weaknesses and tracking sentiment trends over time. Businesses can
identify recurring issues, such as product defects or customer service complaints, and prioritize
improvements accordingly. Furthermore, this analysis facilitates competitive benchmarking by
comparing sentiment data across brands, helping businesses refine their positioning. Sentiment
insights also support reputation management, enabling timely responses to negative feedback
and fostering customer trust.

Despite its benefits, sentiment analysis faces challenges, including handling the nuances
of language, such as sarcasm, context dependence, and cultural diversity in multilingual
datasets. Advances in NLP and artificial intelligence are addressing these complexities, with
newer models offering improved contextual understanding and accuracy. Future developments
in sentiment analysis, such as integrating multimodal data from text, images, and videos,
promise to provide even deeper insights. By harnessing these capabilities, businesses can make
data-driven decisions to enhance product quality, improve customer satisfaction, and maintain a
competitive edge in dynamic markets.

.
 2. INTRODUCTION

Understanding customer sentiment is crucial for businesses in today's digital age, where
online product reviews are a key source of feedback. These reviews offer insights into customer
experiences, preferences, and concerns, providing valuable data for improving products and
services. Sentiment analysis, a technique within natural language processing (NLP), allows
businesses to extract and classify the emotional tone of customer feedback. Among various
approaches, the Naïve Bayes classification algorithm stands out for its simplicity, efficiency, and
effectiveness in analyzing text data. This study focuses on leveraging Naïve Bayes to classify
sentiments in online reviews as positive, negative, or neutral, enabling businesses to better
understand customer opinions.

The process of sentiment analysis involves several essential steps, starting with data
preprocessing. Cleaning raw review data, removing noise, tokenizing text, and normalizing it
ensures that the input data is structured and meaningful for analysis. Using Naïve Bayes, a
probabilistic algorithm based on Bayes' Theorem, reviews are classified into sentiment categories
based on the likelihood of specific words or features appearing in each category. Complementing
this classification, time series analysis tracks sentiment trends over time, offering insights into
how customer perceptions change in response to product updates, campaigns, or market
dynamics.

By implementing this approach, businesses can gain actionable insights into customer
feedback, guiding decisions in product improvement, marketing strategies, and customer
engagement. Tracking sentiment trends enables businesses to address recurring issues, amplify
positive customer experiences, and maintain a competitive edge in a rapidly evolving
marketplace. This analysis highlights the value of using Naïve Bayes for sentiment analysis to
transform customer feedback into a strategic asset.
 3. LITERATURE REVIEW
Sr. No. Title Author Date & Publisher Findings
1. Flipkart B Mohd International Journal Provided actionable insights for
Reviews Munaf, Reyaj of Research Flipkart reviews using Python.
Sentiment Ansari , Syed Publication and Improved sentiment classification
Analysis Omer Ali Reviews, August 2024 with preprocessing and feature
using Khan extraction
Python

2. Sentiment Jitendra Soni, International Journal, Successfully classified Amazon

Analysis of Dr. Kirti Volume 7, Issue reviews into positive, negative, and
Product- Mathur 1,January 2024 neutral sentiments. Addressed
Based emotional polarity using NLP and ML
Reviews techniques
Using
Machine
Learning
Techniques

3. Sentiment Akanksha International Research BERT achieved the highest accuracy

Analysis on Halde, Aditi Journal of (89.25%) for sentiment prediction
Amazon Uttekar, Amit Modernization in compared to other models.
Product Vishwakarma Engineering,
Reviews Technology and
Science,April 2022

4. Sentiment Sharmin International Journal Achieved high accuracy in classifying

Analysis of Shaikh, of Trend in Research Amazon reviews into sentiment
Amazon Aniket and Development, categories using Naive Bayes and
Reviews Navale Volume 7, Issue 6, Logistic Regression
December 2020
 4. PROBLEM STATEMENT

Understanding sentiment from large volumes of online product reviews is a challenging and time-
consuming task when performed manually. The ever-growing amount of customer feedback on e-
commerce platforms, social media, and review sites creates an overwhelming volume of
unstructured text data. This makes it difficult for businesses to efficiently analyze and extract
meaningful insights, as manual efforts are often inconsistent, prone to error, and unsustainable for
large datasets. The complexity of natural language, with its variations in tone, context, and
expressions, further complicates the analysis.

The lack of automated sentiment classification tools capable of handling this volume and diversity
of data poses a significant problem for businesses. Reviews often contain critical feedback on
product performance, usability, and customer experience, which are essential for making informed
decisions. Without an efficient solution to classify sentiments as positive, negative, or neutral,
organizations struggle to identify trends, address customer concerns promptly, and prioritize
improvements. This gap hinders their ability to adapt to customer expectations and remain
competitive in a dynamic market.

Our project addresses this problem by automating the sentiment analysis process. By applying
advanced data analysis techniques, we aim to classify sentiments in online reviews with accuracy
and efficiency. This solution provides businesses with clear insights into customer feedback,
enabling them to enhance product quality, improve customer satisfaction, and make data-driven
decisions. Automation reduces the time and effort required for manual analysis while ensuring
scalability and reliability, empowering organizations to leverage customer feedback as a strategic
asset.
 5. SYSTEM REQUIREMENTS

Technical Requirements
1)Programming Language:
Python:
Selected for its extensive ecosystem, flexibility, and active community support. Offers a variety of libraries
specifically tailored for data preprocessing, NLP, and machine learning tasks. Easy-to-learn syntax ensures
rapid development and experimentation.
Key Libraries and Tools:

(1) Pandas:
Essential for working with structured data, such as CSV files.
Capabilities include:
Data Cleaning: Handle missing values, reformat columns, and filter rows.
Data Manipulation: Grouping, aggregations, and merging datasets.
Data Analysis: Generate descriptive statistics and organize data summaries.

(2) NumPy:
Core library for numerical computations and multi-dimensional array operations.
Key uses include:
Fast mathematical computations.
Efficient handling of large datasets in numerical formats.
NLTK (Natural Language Toolkit):

(3) Comprehensive library for text processing and analysis tasks such as:
Tokenization (breaking text into smaller units like words).
Removing stopwords (e.g., "is," "the," "and").
Stemming (reducing words to their root form, like "running" → "run").
Lexical analysis for feature extraction.

(4) Powerful visualization library used for creating:

Bar charts, histograms, scatter plots, and line graphs.
Enhances the interpretability of data trends and analysis outcomes.
Development Environment:

(5) Jupyter Notebook:

An interactive development environment supporting:
Inline visualizations for immediate feedback.
Easy documentation of the process with embedded markdown.
Step-by-step execution of code for testing and debugging.
Suitable for creating reports and educational presentations
2. Data Requirements

 Data Source:
Online Product Reviews:
Data was collected from Kaggle, a trusted platform providing publicly available datasets.
Focused on real customer reviews of Amazon products to ensure a diverse range of sentiments
and feedback.

 Data Format:

1) CSV (Comma-Separated Values):

A text-based format that stores tabular data, where each row corresponds to a single review and
each column holds specific attributes (e.g., review text, product ID, rating).
Benefits include:
Simplicity and compatibility with most data analysis tools.
Easy integration with Python libraries like Pandas and NumPy.
Key Dataset Features:

2) Review Text:
The primary text of the customer review.
Ratings: Numerical feedback provided by users (e.g., 1-5 stars).
Sentiment Labels: Positive, negative, or neutral sentiments inferred or explicitly provided.
Time of Review: Used for time series analysis to detect sentiment trends.
 6.TECHNOLOGIES USED

Programming Language:
Python
Versatile, high-level programming language well-suited for data analysis, natural language processing, and
machine learning.
Provides a vast library ecosystem for text data processing, modeling, and visualization.

Libraries and Frameworks:

1.Pandas
 For data cleaning, manipulation, and analysis.
 Handles structured datasets efficiently (e.g., CSV files).

2.NumPy
 For fast numerical computations and array manipulations.
 Ideal for large datasets and matrix operations in machine learning.

3.NLTK (Natural Language Toolkit)

 Core library for natural language processing tasks.
 Performs tokenization, stemming, stopword removal, and text normalization.

4.Matplotlib
 Visualization library for creating:
o Histograms, bar plots, scatter plots, and line charts.
 Enables trend and distribution analysis.

5.Naive Bayes Algorithm

 A probabilistic machine learning model used for sentiment classification.
 Chosen for its simplicity and effectiveness in text-based applications.

Development Environment:

· Jupyter Notebook
 Interactive notebook environment that supports live code execution and inline visualizations.
 Combines coding, documentation, and results in one interface.

Dataset Source:

Kaggle

 Trusted platform for publicly available datasets.

 The project utilized Kaggle's Amazon product reviews dataset for analysis.
Visualization and Reporting Tools:

o Word Cloud Generator

 Used to visualize frequently used words in the dataset.

o Power BI or Tableau

 To build interactive dashboards summarizing key insights and trends.

Hardware Requirements:

A system with:

 Minimum 8 GB RAM for handling large datasets.

 Processor: Intel i5 or higher for smooth execution of machine learning tasks.
 Storage: At least 10 GB of free space to store datasets and results.
7. ARCHITECTURE OF PROJECT
 8. MODULES TO BE DEVELOPED

Module 1: Data Collection and NLP (Text Preprocessing)

 Objective: Gather and prepare raw data for analysis.

 Steps Involved:
o Data Collection:
 Sourced Amazon product reviews dataset from Kaggle in CSV format.
 Structured and organized data for further processing.
o Text Preprocessing:

 Missing Value Removal: Eliminated null or incomplete entries.

 Categorical Value Check: Removed irrelevant unique categories.
 Lowercasing: Standardized text by converting all characters to lowercase.
 Punctuation and Noise Removal: Cleared punctuation, symbols, and irrelevant characters.
 Tokenization: Split text into individual tokens (words).
 Stopwords Removal: Removed common stopwords (e.g., "and", "the").
 Stemming: Reduced words to their root forms (e.g., "running" → "run").
 Sentiment Mapping: Translated ratings into sentiment labels (positive, negative, neutral).

Module 2: Exploratory Data Analysis (EDA)

 Objective: Analyze and visualize data for patterns and trends.

 Steps Involved:
o Dataset Overview: Explored structure and summary statistics.
o Sentiment Distribution: Visualized proportions of positive, negative, and neutral sentiments.
o Rating Analysis: Examined customer rating distributions.
o Top Words Analysis: Identified frequently used words with word clouds.
o Visualizations: Created bar charts, histograms, etc., to represent data patterns.
o Correlation Check: Analyzed relationships between ratings and sentiments
 Module 3: Sentiment Classification (Naive Bayes Algorithm)

 Objective: Classify sentiments from preprocessed reviews.

 Steps Involved:
o Algorithm Selection: Chose Naive Bayes for efficient text classification.
o Training: Trained the model using features like word frequencies and sentiment labels.
o Testing: Assessed performance on a separate test dataset.
o Evaluation: Measured model performance through accuracy, precision, recall, and F1-scor
o

Module 4: Time Series Analysis (TSA)

 Objective: Analyze sentiment trends over time.

 Steps Involved:
o Sentiment Trends: Tracked sentiment distribution across time periods (e.g., weeks, months, or years).
o Time-based Visualizations: Created time series plots to visualize sentiment fluctuations.
o Trend Analysis: Analyzed relationships between sentiment changes and events (e.g., product launches).
o Forecasting: Predicted future sentiment trends using historical data

Module 5: Reporting and Insights

 Objective: Summarize findings and provide actionable insights.

 Steps Involved:
o Result Interpretation: Synthesized insights from EDA, TSA, and modeling.
o Actionable Recommendations: Highlighted improvement areas for products or services.
o Final Visualizations: Created concise visual summaries of the analysis.
o Interactive Dashboard: Designed a dashboard using tools like Power BI or Tableau to explore trends
and findings interactively.
9. DATASET COLLECTION
10. CODE
11.OUTPUT
 ADVANTAGES AND DISADVANTAGES
1. Advantages of Sentiment Analysis on Online Product Reviews :

1. Efficient Data Processing:

· Automates the analysis of vast volumes of unstructured text data, saving time and reducing manual
effort.

2.Actionable Insights:

· Provides clear insights into customer opinions, highlighting strengths and areas of improvement in
products and services.

3.Trend Identification:

· Tracks sentiment trends over time, allowing businesses to monitor changes in customer perception after
updates, launches, or marketing campaigns.

4.Improved Decision-Making:

· Facilitates data-driven decisions in product development, marketing strategies, and customer service
enhancements.

5.Customer Satisfaction:

· Identifies recurring complaints and customer pain points, enabling businesses to address issues
proactively and improve satisfaction.

6.Competitive Edge:

· Helps businesses benchmark their products against competitors by comparing sentiment trends across
brands.

7. Scalability:

· Can handle increasing volumes of data as businesses grow, making it a sustainable solution for ongoing
feedback analysis.

8. Cost-Effective:

· Reduces the need for large teams dedicated to manually reviewing feedback, cutting operational costs.
 2.Disadvantages of Sentiment Analysis on Online Product Reviews:
1.Contextual Limitations:

· Struggles with understanding nuances like sarcasm, irony, and ambiguous language, which can lead to
misclassification.

2.Linguistic Diversity:

· Difficulty in analyzing reviews written in multiple languages or containing slang, colloquialisms, and
regional dialects.

3.Quality of Data:

· Results are heavily dependent on the quality of input data, which may include noise, irrelevant
information, or incomplete reviews.

4. Bias in Analysis:

· Algorithms may inherit biases present in the training data, leading to skewed results and inaccurate
insights.

5.Interpretation Challenges:

· Requires skilled analysts to interpret sentiment analysis results correctly and link them to actionable
business strategies.

6.Privacy Concerns:

· Collecting and analyzing online reviews might raise ethical and privacy issues, especially if the
reviews contain identifiable personal data.

7.Real-Time Analysis Complexity:

· Analyzing reviews in real time can be resource-intensive and challenging to implement effectively for
businesses dealing with high-frequency data.

8.Dependency on Preprocessing:

· Sentiment analysis heavily relies on effective data preprocessing, which can be time-consuming and
requires domain knowledge.

.
 14. CONCLUSION

Sentiment analysis on online product reviews offers a powerful tool for businesses to understand
customer opinions, track trends, and gain actionable insights into product performance. By
automating the process of classifying sentiments as positive, negative, or neutral, businesses can
efficiently process large volumes of unstructured data and make informed decisions quickly. The
ability to monitor sentiment trends over time further enhances the capacity to respond to customer
needs, address concerns, and improve products or services, ultimately boosting customer
satisfaction and loyalty.

However, while sentiment analysis provides valuable benefits, it also comes with challenges. Issues
such as handling linguistic nuances, context-specific interpretations, and the quality of the data can
affect the accuracy and reliability of results. Additionally, concerns related to biases in the analysis
and privacy issues must be addressed to ensure ethical and fair outcomes. Despite these challenges,
ongoing advancements in natural language processing and machine learning continue to improve
the capabilities and accuracy of sentiment analysis.

In conclusion, sentiment analysis on online product reviews is an invaluable tool for businesses
seeking to remain competitive in a customer-driven marketplace. By leveraging the power of data
and technology, companies can gain deeper insights into customer feedback, refine their strategies,
and make data-driven decisions that lead to enhanced products and better customer experiences. As
technology continues to evolve, the effectiveness and scope of sentiment analysis will only continue
to expand, offering even more opportunities for businesses to connect with their customers and
drive success.
 15. BIBLIOGRAPHY

· Sentiment Analysis of Product Reviews Using Natural Language Processing

Journal of Computer Science and Technology, Volume 12, Issue 2, March 2023
This paper explores the use of Natural Language Processing (NLP) techniques for sentiment analysis on
product reviews. It delves into various preprocessing steps, such as tokenization and stopword removal,
and how machine learning models can classify customer feedback effectively.

· A Comparative Study of Sentiment Analysis Techniques for E-Commerce Product Reviews

International Journal of Data Science and Artificial Intelligence, Volume 9, Issue 4, September 2023
This comparative study evaluates different sentiment analysis techniques, including Naïve Bayes, Support
Vector Machines (SVM), and deep learning approaches, for analyzing e-commerce product reviews. It
highlights the strengths and weaknesses of each method in the context of real-world online product
feedback.

· Enhancing E-Commerce Business Intelligence with Sentiment Analysis on Product Reviews

International Journal of Business Analytics, Volume 15, Issue 2, February 2022
This paper focuses on how sentiment analysis of online product reviews can be integrated into business
intelligence systems to provide insights into customer preferences, improve product development, and
enhance customer engagement strategies for e-commerce businesses.

· Flipkart Reviews Sentiment Analysis using Python

International Journal of Research Publication and Reviews, August 2024
This paper focuses on applying sentiment analysis to Flipkart product reviews using Python and machine
learning techniques. It explores how sentiment can be classified and used to improve customer
experience on e-commerce platforms.

· Sentiment Analysis of Product-Based Reviews Using Machine Learning Techniques

International Journal, Volume 7, Issue 1, January 2024
This study investigates various machine learning techniques for sentiment analysis of product reviews.
The paper highlights different algorithms for classifying customer feedback and their applications in
enhancing business decision-making.

· Sentiment Analysis on Amazon Product Reviews

International Research Journal of Modernization in Engineering, Technology and Science, April 2022
This research focuses on the sentiment analysis of product reviews on Amazon. It uses machine learning
and NLP techniques to evaluate customer sentiments and provide insights into product improvement and
customer satisfaction.

· Sentiment Analysis of Amazon Reviews

International Journal of Trend in Research and Development, Volume 7, Issue 6, December 2020
The paper discusses sentiment analysis of Amazon reviews and how businesses can use sentiment
classification to understand customer opinions. It explores the effectiveness of different sentiment
analysis techniques and their impact on product marketing strategies.