LABORATORY REPORT
NAME: Manish Gupta
SEC: B ROLL: 22CS011090
SUB: Software Engineering Lab
SUB CODE: YCS6102
Page | 1
ACKNOWLEDGEMENT
I would like to express my sincere gratitude to Dr. Radhakrishna Jana,
whose guidance and support have been instrumental in the completion of
this Environmental Science assignment. Her insightful feedback and
encouragement have greatly enhanced my understanding of the subject.
I would also like to thank JIS University for providing an excellent learning
environment and resources that have been crucial for my academic
growth. The knowledge and skills I have acquired here have been
invaluable.
Date: 02.05.2025
Place: JIS University
Page | 2
INDEX
Seri
al Experiment Title Date Signature
No.
Problem Statement 17/02/
1
Definition 25
2 17/02/
Feasibility & SRS Development
25
24/02/
3 DFD & Structured Chart
25
24/02/
4 Use Case Analysis
25
Class & Object 03/03/
5
Diagrams 25
State & Activity 03/03/
6
Diagrams 25
Sequence & 10/03/
7
Collaboration Diagrams 25
Component Diagram 10/03/
8
Implementation 25
31/03/
9 Deployment Diagram Design
25
Unit & Integration 07/04/
10
Testing 25
Effort Estimation 14/04/
11
Techniques 25
21/04/
12 Project Planning Charts
25
Page | 3
01. Write down the problem statement for a suggested system
of relevance. [1 day]
PROBLEM STATEMENT
The primary challenge addressed by a face mask detection system is
the need for an automated, real-time solution to monitor and enforce
mask compliance in diverse environments such as public spaces, offices,
schools, and transportation hubs. Traditional manual methods of
monitoring mask usage are often impractical and inefficient, especially
in high-traffic areas with large populations. Therefore, a system capable
of automatically detecting the presence or absence of face masks on
individuals becomes a valuable tool for maintaining a safe and healthy
environment.
The goal is to predict whether the person is wearing a mask or not.
Since this is an issue of binary classification, the model if predicts
correctly then a green rectangular box appears on person wearing
mask.
If the person is not wearing mask, the model predicts red rectangular
box on his face along with accuracy score.
TITLE: Face Mask Detection Using Machine Learning
Page | 4
02. Do Feasibility study along with requirement analysis and
develop Software Requirement Specification Sheet (SRS) for
suggested system. [1 day]
FEASIBILITY STUDY
The development of a face mask detection system using machine
learning has been demonstrated to be technically feasible through
numerous research endeavors. Studies employing various machine
learning and deep learning techniques, including Convolutional Neural
Networks (CNNs) and architectures like MobileNetV2, have reported high
accuracy rates, ranging from 92.53% to 99.98% on training datasets.
This indicates that the underlying algorithms are capable of learning and
generalizing the features necessary for accurate mask detection.
Furthermore, the feasibility of real-time detection on video streams
using libraries such as OpenCV has been widely documented, making
the technology applicable to live monitoring scenarios.
The selection of the appropriate deep learning model plays a crucial role in
balancing the trade-off between accuracy and performance, a key factor in
the feasibility assessment. For instance, MobileNetV2 is often favored for its
efficiency and suitability for real-time applications, while models like Faster R-
CNN might offer higher accuracy at the cost of increased computational
demands. The widespread availability and support for open-source libraries
such as OpenCV, TensorFlow, Keras, and PyTorch significantly contribute to
the feasibility by simplifying the development and deployment processes,
providing a robust ecosystem of tools and resources.
Reporte
Computatio
d
nal
Accurac Key
Model Resource
y Characteristics
Requireme
(Trainin
nts
g Data)
Efficient, suitable
92.53%
MobileNet for mobile and
- Low
V2 embedded
99.98%
systems
81%
(Averag Potentially higher
Faster R-
e High accuracy for
CNN
Precisio object detection
n)
Real-time object
detection,
YOLOv5 95% Medium
balance of speed
and accuracy
Combines SSD
98% Page | 5
for face
CL- (Test
- detection and
SSDXcept Accurac
Exception for
y)
classification
Requirement Analysis and Software Requirement Specification (SRS)
The face mask detection system requires a set of functional and non-functional
capabilities to meet its objectives.
Functional Requirements:
1. Face Detection: Accurately detect human faces in images/video streams.
2. Mask Classification: Classify each detected face as "Mask" or "No
Mask" with high accuracy.
3. Real-Time Processing: Process video streams in real-time (e.g., ≥15 FPS)
for immediate feedback.
4. Multi-Mask Support: Handle variations in mask types, colors, and designs.
5. Alert System: Generate alerts (visual/sound/SMS/email) for non-
compliance.
Non-Functional Requirements:
1. Accuracy: Minimize false positives/negatives (target >95% precision).
2. Robustness: Reliable performance under varying lighting, angles, and
occlusions.
3. Performance: Fast inference speed (e.g., <100ms per frame with GPU).
4. Reliability: Consistent results over prolonged usage.
5. Maintainability: Modular design for easy updates and scalability.
Category Requirement Specifications
Operating System Platform compatibility Windows 10/11, Linux
(Ubuntu 20.04+)
Programming Development language Python 3.8+
Language
Libraries Essential ML/CV TensorFlow 2.x, Keras,
frameworks OpenCV 4.5+, PyTorch
(optional)
CNN Model Detection/Classification MobileNetV2
model (pretrained), ResNet50
(custom-trained)
Processor (CPU) Minimum processing Intel i5 8th Gen / AMD
power Ryzen 5 (4+ cores)
RAM System memory 8GB (16GB
recommended)
Storage Disk space for 256GB SSD (500GB+ for
datasets/models large datasets)
GPU (Recommended) Accelerated deep NVIDIA GTX 1660 / RTX
Page | 6
learning 3060 (6GB+ VRAM)
Camera Input resolution & FPS 1080p (1920x1080) @
15-30 FPS
03. To perform the function-oriented diagram: Data Flow
Diagram (DFD) and Structured chart. [1 day]
1. Level 0 or Context Diagram:
2. Level 1:
3. Level 2:
Page | 7
04. To perform the user’s view analysis for the suggested
system: Use case diagram. [1 day]
Page | 8
05. To draw the structural view diagram for the system: Class
diagram. [1 day]
Page | 9
06. To draw the behavioral view diagram: Activity diagram. [1
day]
Page | 10
07. To perform the behavioral view diagram for the suggested
Page | 11
system: Sequence diagram and State diagram. [1 day]
Sequence diagram
State Diagram
Page | 12
08. To perform the implementation view diagram: Component
diagram for the system. [1 day]
Page | 13
09. To perform the environmental view diagram: Deployment
diagram for the system. [1 day]
Page | 14
10. To perform various testing using the testing tool unit
testing, integration testing for a sample code of the suggested
system. [1 day]
Page | 15
Test
Test Component Expected Statu
Case Test Description
Type / Feature Result s
ID
Image Test camera input Image is
Pendi
UT001 Capture and image captured and
ng
Module capture. saved.
Image Image is
Verify resizing and Pendi
UT002 Preprocessin preprocessed
normalization. ng
g correctly.
Face bounding
Unit Face Detect faces in Pendi
UT003 boxes are
Testin Detection given test images. ng
accurate.
g
Test ML model Correct
Mask Pendi
UT004 prediction with mask/no-mask
Classification ng
known images. label returned.
Alert Verify alert Alert is triggered Pendi
UT005
Generation creation logic. when no mask. ng
Logging Ensure detection Data is saved in Pendi
UT006
Module logs are recorded. log format. ng
11. Perform Estimation of effort using FP Estimation for
chosen system with other matrices. [1 day]
Function Point Estimation:
To perform FP estimation, we need to identify the five key components:
● External Inputs (EI): Data entering the system from the user or external
systems (e.g., video stream). Let's assume 1 complex EI (continuous video
feed).
● External Outputs (EO): Data leaving the system to the user or external
systems (e.g., alerts, statistics). Let's assume 2 simple EOs (visual alert, log
entry).
● External Inquiries (EQ): Requests from the user that require the system
to retrieve data (e.g., request for statistics - optional use case). Let's
assume 1 average EQ.
● Internal Logical Files (ILF): Data stored within the system (e.g., trained
model). Let's assume 1 complex ILF (the trained ML model).
● External Interface Files (EIF): Files accessed by the system but
maintained by other systems (not directly applicable in this scenario as
described). Let's assume 0 EIF.
Page | 16
Now, we assign complexity weights (low, average, high) to each component
based on standard FP guidelines.
Component Quantity Complexity Weight FP
Type Contribution
External 1 Complex 6 6
Inputs (EI)
External 2 Simple 4 8
Outputs (EO)
External 1 Average 4 4
Inquiries (EQ)
Internal 1 Complex 15 15
Logical Files
(ILF)
External 0 N/A N/A 0
Interface Files
(EIF)
Unadjusted 33
FP (UFP)
Page | 17
12. To prepare time line chart/Gantt Chart/PERT Chart for
selected software project. [1 day]
Page | 18