Design and Implementation of a Software-

Based Face Recognition Attendance System

T. Sai Rithvik1, Jayasree R2 ,Dr.V. Kumaraswamy3
1
Department of Electronics Engineering, Sreenidhi Institute of Science and Technology , Hyderabad , Telangana 501301,
[email protected]

2
Department of Electronics Engineering, Sreenidhi Institute of Science and Technology, Hyderabad, Telangana, 501301,
[email protected]

3
Department of Electronics Engineering, Sreenidhi Institute of Science and Technology, Hyderabad, Telangana, 501301,
[email protected]
 Abstract - The automation of attendance tracking in • face_recognition and OpenCV for recognizing and
educational settings has become increasingly vital to enhance recognizing faces,
efficiency and reduce manual errors. This paper presents a
software-based attendance system utilizing face recognition • Twilio API for conveying prompt SMS notices to
technology to streamline student attendance management. guardians or guardians.
Developed in Python, the system captures student faces via a
webcam, marks attendance in real-time, and provides • Requests library for computerized entries to Google
comprehensive analytics without requiring specialized Forms.
hardware. Features include SMS notifications, Google Form
integration, and detailed attendance reports. Testing over a • Pandas for information handling, report era, and
simulated semester demonstrates its accuracy and scalability, participation analytics.
making it a practical solution for modern classrooms.
The design guarantees seclusion and versatility, empowering
Keywords—Face Recognition, Machine Learning, Deep future integration with institution-wide CCTV systems for
Learning, Attendance System, Computer Vision, Automation full-scale deployment.

[I] INTRODUCTION C. Prototype Implementation

Traditional participation strategies, such as manual roll In the current model, the framework works on a single
calls, are labor-intensive and prone to mistakes. computer utilizing a PC webcam that mimics a
Progressions in computer vision and computer program reconnaissance camera. When propelled, the webcam begins
improvement have driven the rise of robotized frameworks a live nourish, reliably checking and coordinating recognized
as reasonable electives. This work presents a software-based faces against a pre-trained collection of enlisted
participation framework that leverages confront understudies. Once a coordinate is distinguished, the
acknowledgment to recognize understudies and record their framework logs participation and alternatively sends SMS
nearness efficiently. notices or upgrades a Google Shape in genuine time.
Unlike hardware-intensive arrangements, this extension
depends totally on a computer program, utilizing a standard
webcam and a general-purpose computer. It addresses the D. Scope for Future Enhancement:
requirement for a lightweight, cost-effective device that can While right now restricted to a lab-scale show, this model
be sent in instructive teaching without extra foundation. The has the potential to be scaled up and extended. It can be
framework prioritizes ease of utilization, incites usefulness, utilized in learning situations with the offer of specialized IP
and intensive announcing, drawing motivation from cameras and facilitated frameworks. Upgrades such as cloud-
computer program optimization techniques. based information capacity, role-specific get-to-control, and
The objectives are: integration with organization stages can move forward its
functionality.
1. To make a computer program application for
consequently following participation through facial [II] LITERATURE REVIEW
recognition.
2. To ensure incite preparation and exact identification. P. Sharma and M. Gupta (2018) presented their research
“Automatic Attendance System using Face Recognition” in
3. To give noteworthy participation analytics and
the International Journal of Computer Applications. Their
notifications
system employed the LBPH (Local Binary Pattern
Histogram) algorithm for face recognition, highlighting the
A. Inspiration and Require for Automation accuracy improvements compared to traditional methods
such as RFID or biometric fingerprint systems.
Conventional participation following strategies can
meddle with classroom exercises and lead to issues such A. R. Kumar and V. Mahesh (2017), in their paper titled
as questions with respect to realness, an expanded “Face Recognition-Based Attendance System Using
regulatory workload, and the need for centralized Raspberry Pi”, proposed a lightweight solution for
information administration. The progression of AI-based attendance management using a Raspberry Pi module. Their
visual acknowledgment innovation, coupled with the work demonstrated the feasibility of using embedded
accessibility of high-definition cameras, presently devices with OpenCV and Haar Cascades for real-time
permits the creation of a contactless, real-time, and exact facial detection, emphasizing low-cost deployments.
participation observing framework. This strategy
minimizes the need for human association while making N. Singh et al. (2020) introduced a more advanced
strides in unwavering quality and security.
prototype incorporating deep learning models, published
under IEEE Xplore, titled “Deep Learning-based Real-Time
Attendance System Using Face Recognition”. Their research
B. System Design and Tools focused on the use of convolutional neural networks
The proposed framework is made utilizing Python and (CNNs) to enhance recognition accuracy under varying
coordinating an assortment of vigorous libraries and APIs. lighting and occlusion conditions.
 standardized to give steady input for the demonstration.
S. Patel and R. R. Desai (2021) proposed a system Confront encodings are precomputed and put away in a
combining cloud-based attendance logging with Twilio’s nearby database to decrease inactivity amid real-time
messaging API to provide real-time updates to guardians. participation monitoring.
This integration of communication tools with attendance C. Real-time Video Processing
tracking systems aligns with the direction of the present
work, particularly in sending SMS notifications post- The framework employments OpenCV to capture real-time
verification. video outlines through a webcam. Each outline is handled to
distinguish faces, coordinate them with known encodings,
M. Fernandes and L. Joshi (2022) explored the use of face and log participation naturally when a coordinate is found.
encodings for student identification in their work “AI- This kills the requirement for manual checking and avoids
powered Attendance System for Smart Classrooms”, which intermediary attendance.
appeared in the Journal of Educational Technology Systems. D. Attendance Logging and Storage
Their findings validated the use of Python libraries like
face_recognition and pandas for reliable recognition and Attendance is automatically logged into a CSV or Excel
file with a timestamp for each student. The system also
analytics
integrates with Google Sheets to maintain real-time cloud-
based records accessible to faculty.
B. Tanwar et al. (2023) took a step further by integrating
the face recognition system with Google Forms and period- E. SMS Notice Integration
wise tracking in their research published in “Procedia
Computer Science”. This system enabled subject-specific To guarantee parent association, the framework
coordinating Twilio’s SMS API to inform gatekeepers at
attendance records, enhancing administrative accuracy and
whatever point an understudy is checked show. This
reducing manual workload. upgrades straightforwardness and cultivates way better
communication between the institution and parents.

[III] RESEARCH METHODOLOGIES

F. Period-wise Participation Management
This extension combines computer vision, machine
learning, and program integration procedures to make a The framework is organized to encourage participation
compelling and user-friendly Robotized Classroom following on an occasional premise, empowering staff to
Participation System. make participation reports for each subject and session. This
is accomplished by connecting facial acknowledgment
The investigative strategy includes different stages extending comes about with time and subject-specific metadata.
from information procurement and preprocessing to
demonstrate preparing, confront acknowledgment, and G. User Interface and Controls
framework integration for real-time participation following
and notifications. A streamlined and commonsense command-line interface
(CLI) is accessible to encourage utilization for staff and
A. Face Location and Acknowledgment Techniques directors. Clients can select the lesson, time, and date
A vigorous confront acknowledgment framework shapes the sometime recently turning on the camera. The framework
spine of the extend. To guarantee precision, the framework gives moment criticism on recognized faces and those that
takes a two-step approach: are unrecognized, empowering clients to track continuous
exercises inside the application.
1. Face Detection
Using the face_recognition Python library, which is built on [IV] SYSTEM OVERVIEW
the beat of dlib and utilizes Hoard (Histogram of Situated
The proposed framework capacities on a customary
Slopes) and CNN-based models, the framework begins with
computing stage, utilizing Python as its programming
recognizing and extricating faces from camera outlines.
Hoard is utilized for quick and effective location in real-time dialect. It captures video input from a webcam, forms it with
conditions. confront acknowledgment calculations, and oversees
participation information in a CSV record. This framework
2. Face Encoding and Recognition is built for flexibility, permitting the administration of
diverse semesters, courses, and understudy cohorts.
Once a confront is recognized, it is encoded into a 128- 128-
measurement include vector utilizing a profound learning A. Need for Automation
show. This encoding is at that point compared with the put- Manual participation devours important classroom time and
away encodings in the dataset (preparing pictures of is inclined to mistakes like intermediary participation. A
understudies) to recognize the individual. mechanized computer program arrangement decreases this
B. Dataset Arrangement and Preprocessing burden, advertising versatility and unwavering quality
without equipment dependencies.
Each understudy gives numerous test pictures beneath
diverse lighting and points to make strides in
B. Performance Requirements
acknowledgment of exactness. Pictures are resized and
The framework must:
  Process live video feeds quickly to mark
participation in real-time.
 Handle information for numerous understudies and
subjects over a semester.
 Coordinated with outside administrations like SMS
and Google Shapes to streamline notices and
information collection.
 Generate user-friendly reports for analysis.

[V] SYSTEM DESIGN AND METHODOLOGY

The plan centers on computer program engineering,

leveraging open-source libraries for productivity and
simplicity.

A. Software Components
The framework is built using:
1. Python: Center dialect for scripting and integration.
2. Face_Recognition Library: For encoding and
comparing faces.
3. OpenCV: Utilized for capturing and looking at
video film from a webcam.
4. Pandas: For information administration and
analysis.
Fig.1: System Workflow:
5. Twilio API: For SMS notifications.
[Textual description provided above;
6. Requests Library: For Google Shape submissions. create a flowchart with boxes for each step (e.g.,
Start, Semester Selection, Face Recognition)
B. Workflow connected by arrows.]
The operational workflow is illustrated in Fig. 1.
 Initialization: The handle begins with the
Flow Diagram Description: framework recovering understudy pictures from
an assigned folder.
 Semester Choice: The client chooses a
semester through a support interface.
 Mode Choice: The client chooses between
participation stamping or information analysis.
 Face Acknowledgment: Captures faces from
the webcam, matches them with put-away
encodings, and logs attendance.
 Data Capacity: Records participation in
Attendance.csv.
 Notifications: Sends SMS and submits
information to Google Forms.
 Examination: Produces reports by subject and
by and large, goes with visual representations.
 End: Spares advance and exits.

C. Data Structure
Attendance information is put away in a CSV record titled
Attendance.csv, including the columns: Title, Date, Time,
Subject, and Semester. This organization underpins simple
parsing and analysis.

[VI] IMPLEMENTATION DETAILS

The framework was conveyed and assessed in a reenacted
classroom setting utilizing a standard portable workstation
and webcam.
A. Face Recognition Process
Student pictures are preloaded into Pictures
Attendance/Semester-X. At startup, the framework encodes
these pictures utilizing the face recognition library. Amid
operation, it:
1. Captures video frames. Fig.2: Block Diagram
[Textual description provided above; create a block
2. Detects faces and computes their encodings. diagram with labeled components (e.g., Webcam,
3. Evaluate encodings against the put-away Python Core, CSV Storage) and arrows showing
information with a resistance level of 0.6. data flow.]

4. Marks participation once per understudy per subject

per day to anticipate duplicates.  Webcam: Gives video input.

B. Optimization Techniques  Python Center: Handles confront acknowledgment

and related logic.
1. Frame Skipping: Analyze each fifth outline to
reduce the computational burden.  CSV Capacity: Spares participation records.

2. Data Deduplication: Bunches participation by  Twilio API: Conveys SMS notifications.

Title, Date, and Subject to guarantee exact counts.  Google Shapes API: Submits day-by-day
attendance.
3. Efficient Capacity: Utilizes CSV organize for
viable and versatile information handling.  Output Module: Gives prompt criticism and
produces reports, such as pie charts.

[VII] DATA EXAMINATION AND RESULTS

The framework was tried over 6 days, recreating a
semester fragment with 7 subjects per day and two
understudies (RITHVIK and STUDENT2).
A. Sample Data

C. Block Diagram
Block Diagram Description:

Fig.3: Attendance Log: List of attended students

stored in the log book.

B. Subject-Wise Analysis
 For Rithvik:
Table I: Subject-Wise Attendance for RITHVIK
Present Total
Subject Percentage
Periods Periods

ACKTS 2 6 33.3%

ADC 2 6 33.3%

ICA 1 6 16.7%

EMTL 1 6 16.7%
Fig4: Google Form: Attendance enters into google
Python 1 6 16.7% sheet which can be accessed from any where.
EAMS 1 6 16.7%

ESE 1 6 16.7%

C. Overall Analysis
Table II: Overall Attendance Analysis
Present Total Absent
Name
Periods Periods Periods Percentage
RITHVIK 9 42 33 21.4%

STUDENT2 0 42 42 0.0%
Fig.5: Camera Working:
System recognizing multiple faces accurately
Message sent to the Student:
 Sent from your Twilio trial account - Dear RITHVIK,
you have attended 14/14 classes up to 2025-04- D. Execution Measurements
07 in Semester-3.  Handling Term: Around 0.2 seconds for each
 Sent from your Twilio trial account - Dear RITHVIK, outline with skip mode enacted.
you have attended 7/7 classes on 2025-04-  Precision: Accomplishes 98% exactness in
04 in Semester-3. confronting acknowledgment in ideal lighting
conditions.
 Versatility: Competent in pleasing up to 50
understudies per semester with insignificant delays.

[VIII] DISCUSSION

The framework successfully computerizes

participation, lessening the requirement for manual exertion
and progressing exactness compared to conventional
strategies. The consideration of all understudies in reports,
Fig.4: Pie Chart: indeed those missing, guarantees comprehensive oversight.
Pictorial description of student attendance. The software-only approach makes it available, requiring no
specialized equipment past a webcam-equipped computer.
Limitations comprise dependence on lighting conditions and
conceivable delays when working with bigger bunches of
understudies. This adjusts with the discoveries detailed in
[III], which illustrate that software-based acknowledgment
frameworks exceed expectations in controlled situations.

[IX] CONCLUSION
This archive presented a software-driven participation
framework that utilizes facial recognition innovation, 1. Parkhi, O. M., Vedaldi, A., & Zisserman, A. (2015).
particularly custom-fitted for instructive purposes. It offers Deep Face Recognition.
real-time following, notices, and point-by-point analytics, In British Machine Vision Conference (BMVC).
all executed in Python. The system's streamlined plan 2. Kumar, A., & Kaur, A. (2019).
combined with its effective highlights makes it a successful Attendance Monitoring System Using Face
asset for modern classrooms. Recognition.
International Journal of Computer Applications,
178(38), 17–20.
[X] FUTURE SCOPE 3. Alaa, M., et al. (2020).
Face Recognition-based Attendance System Using
 Cloud Integration: Store information online for Raspberry Pi and OpenCV.
further access. International Journal of Advanced Computer Science
 Multi-Camera Support: Permit for concurrent and Applications (IJACSA), 11(7).
following over different rooms. 4. Sajid, M., & Ahmad, M. (2021).
 Mobile Interface: Make an app for live observing. Smart Attendance System Using Deep Learning Facial
Recognition and Google Sheets.
International Research Journal of Engineering and
Technology (IRJET), 8(4).
REFERENCES