International Journal for Multidisciplinary Research (IJFMR)

E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: editor@ijfmr.com

Smart Attendance System: Using Facial

Recognition Technology
Amaan Ansari1, Anita Pujari2, Dr. Kamika Chaudhary3
1,2
Master of Computer Science, M.B.G.P.G. College Haldwani, Kumaun University Nainital (263139)
3
Assistant Professor, Department of Computer Science, M.B.G.P.G. College Haldwani, Kumaun
University Nainital (263139)

Abstract:
The Smart Attendance system automates attendance tracking using advanced face recognition technology
integrated with cloud-based data storage. This system includes components for generating face encodings,
populating a database with student details, and tracking attendance in real time through live video feeds.
The face encoding generation processes student images to create unique facial encodings stored alongside
student IDs, Real-time attendance tracking captures live video, detects faces, and matches them against
stored encodings to update attendance records in the Firebase Real-time Database. This paper describes a
method for Student Attendance system that will record the attendance of the students in the classroom
environment automatically and it will provide the facilities to the faculty to access the information of the
students easily.

Keywords: Face recognition, python, pycharm, libraries, firebase, opencv, database.

Introduction:
Attendance tracking is essential in educational institutions and workplaces, yet traditional methods like
manual sign-ins and roll calls are inefficient and prone to inaccuracies and manipulation. This research
explores a Face Recognition Attendance System to overcome these challenges, leveraging the unique
characteristics of human faces for secure, contactless, and accurate identification. The system comprises
three main components: face encoding generation, database population, and real-time attendance tracking.
Face encoding generation processes student images to create unique facial encodings stored with student
IDs, managed by the encodegenerator.py script, which also uploads images to Firebase Storage. The
database population module, implemented in addtodatabase.py, initializes the Firebase Real-time
Database with student details, ensuring necessary information is available for attendance tracking. The
real-time attendance tracking module, detailed in main.py, captures live video feeds, detects faces, and
matches them against stored encodings to update attendance records in the Firebase Real-time Database.
Additional functionalities include automatic data ingestion from external sources and the generation of
monthly attendance reports in Excel format. Extensive testing has demonstrated the system's high accuracy
and reliability under various conditions, including changes in lighting and occlusions. Integration with
Firebase ensures real-time data synchronization and scalability, making the system suitable for
deployment in educational institutions and other organizational settings. The objectives of this study
include developing a robust facial recognition algorithm capable of handling diverse conditions, ensuring
data privacy and security, integrating the system

IJFMR240323705 Volume 6, Issue 3, May-June 2024 1

 International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: editor@ijfmr.com

with existing attendance management systems, and evaluating its performance and user acceptance. The
system aims to be scalable and cost-effective, providing significant administrative efficiencies compared
to traditional methods.

Literature Review:
Face recognition has been a significant area of research within computer vision, driven by both its practical
applications and theoretical interest from cognitive scientists. While other biometric identification
methods, such as fingerprint or iris scans, may offer higher accuracy, face recognition remains a primary
focus due to its non-invasive nature and because it mirrors the primary way humans identify one another.
The technology behind face recognition is steadily advancing towards becoming a universal biometric
solution, mainly because it requires minimal effort from users compared to other biometric options.
Biometric face recognition is predominantly used in three main areas: time and attendance systems, visitor
management systems, and access control systems.
Traditionally, student attendance is recorded manually using attendance sheets provided by faculty
members in class. This method is time-consuming and inefficient, especially in large classrooms with
numerous students, making it challenging to verify each student individually. The authors of this paper
propose a system that uses face recognition to automate attendance recording, ensuring that the presence
of enrolled individuals is accurately logged within the respective venue. The proposed system also
maintains a log file to keep records of each individual's entry based on a universal system time.
Early efforts in face recognition began in the 1960s with semi-automated systems where key facial features
were marked on photographs, and distances and ratios were calculated and compared to reference data. In
the early 1970s, Goldstein, Harmon, and Lesk [1] developed a system using 21 subjective markers such
as hair color and lip thickness. This method proved challenging to automate due to the subjective nature
of the measurements, which were performed manually.
Fisher and Elschlager [2] developed methods to measure different facial features and map them onto a
global template, but these features did not provide enough unique data to represent an adult face accurately.
Another method, the Connectionist approach [3], attempted to classify human faces using a combination
of gestures and identifying markers, typically implemented using 2D pattern recognition and neural
network principles. However, this approach required many training faces to achieve reasonable accuracy,
limiting its large-scale implementation.
The first fully automated face recognition system [4] was developed based on general pattern recognition
principles. It compared faces to a generic face model of expected features and generated patterns for an
image relative to this model. This statistical approach relied on histograms and grayscale values, forming
the foundation for future advancements in face recognition technology. [5]

Methodology:
This research paper details how face recognition can be effectively employed to automate attendance
tracking, significantly improving accuracy and efficiency in educational settings. The proposed system
not only automates attendance recording but also ensures real-time updates and maintains a log for
verification purposes, highlighting its potential to streamline administrative tasks and enhance overall
management.
The system comprises several scripts, each with specific responsibilities. The encodegenerator.py script
handles loading images, detecting faces, generating encodings, and saving them to a database or file. The

IJFMR240323705 Volume 6, Issue 3, May-June 2024 2

 International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: editor@ijfmr.com

addtodatabase.py script manages adding new face images or encodings to the database, ensuring no
duplicates, and updating the database with new entries. The main.py script captures real-time video feeds,
detects and recognizes faces, and logs attendance for recognized faces. A system flow diagram outlines
the sequence of operations: initialization with encoding generation and storage, adding new data via the
database script, and tracking attendance through real-time video capture and recognition. The
programming languages and libraries used include Python for core functionality, JavaScript for a web-
based interface, OpenCV for video processing, and various other Python libraries for data manipulation,
Firebase interaction, and report generation. The Face Recognition Attendance System utilizes a modular
architecture to ensure efficient and accurate attendance tracking. This modular approach supports
scalability, maintainability, and ease of development, with each component playing a critical role in the
overall functionality, contributing to a robust and automated solution for attendance management.

Implementation and Data Collection:

The project was developed using PyCharm IDE, chosen for its robust features supporting Python
development, including code completion, debugging tools, and version control integration.
Key technologies such as Python for its simplicity and extensive library support, face recognition for face
detection and recognition, opencv-python for image processing, and firebase-admin for database
integration were pivotal in achieving real-time attendance tracking with face recognition. The
implementation started by setting up Firebase credentials and configuring the video capture device.
Precomputed face encodings were loaded from files to facilitate efficient face recognition. The core
functionality involved capturing video frames, detecting faces using face recognition, comparing face
encodings, and updating attendance records in Firebase based on recognized faces.

Image Collection:
Images are sourced from a local directory named 'images', where each image corresponds to a unique
student identified by their student ID. This structured approach facilitates easy access and management of

IJFMR240323705 Volume 6, Issue 3, May-June 2024 3

 International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: editor@ijfmr.com

student data.

Firebase Storage Integration:

Collected images are uploaded to Firebase Storage, ensuring centralized and secure storage. This
integration enables seamless access to images across various components of the system. Face encoding
extracts numerical representations of facial features crucial for recognition. Leveraging the face
recognition library, each image undergoes processing to generate these encodings. The find Encodings
function converts images to RGB format and computes corresponding face encodings, which are stored
for subsequent recognition tasks. Integration with Firebase encompasses leveraging Firebase Real-time
Database and Firebase Storage for effective data management. Firebase is initialized with appropriate
credentials, facilitating secure interactions with Firebase Real-time Database and Storage components.
Images collected from local directories are uploaded to Firebase Storage, establishing a centralized
repository accessible to all system modules. This centralized approach simplifies image management and
enhances scalability.

Database Design, Testing and Results:

Database Schema Design outlines the structured representation of student information and attendance
records. Each student is uniquely identified by a Student ID and associated with attributes such as name,
subject, starting year, total attendance count, current year and last attendance timestamp. This structured
schema facilitates efficient data organization and retrieval.
Integration with Firebase Real-time Database involves initializing Firebase using the Admin SDK and
configuring database access through a service account key. This setup ensures secure authentication and
enables seamless interaction with Firebase services. Storing Data utilizes Firebase's set method to write
student information into the database under their respective Student IDs. Data is structured as dictionaries,
ensuring clarity and accessibility of stored information. Retrieving Data involves using Firebase's get
method to fetch specific student information from the database based on their Student ID. This real-time
retrieval capability supports dynamic updates and verification of attendance records during face
recognition processes. Real-Time Applications leverage the retrieved data to dynamically verify
attendance and update records as needed. For instance, the system can automatically update total
attendance counts and timestamps based on real-time face recognition results.

Database Integration:
Firebase Real-time Database is employed as the backend solution for storing attendance records due to its
real-time data synchronization capabilities and scalability. Firebase ensures immediate updates to
attendance records as faces are detected and recognized, thereby maintaining accurate and current
attendance data. Security features like Firebase Authentication are utilized to restrict access to authorized
personnel only, ensuring data integrity.

Single Instance Attendance:

The system utilizes single face detection and recognition techniques where each student registers their
face through images for training the facial recognition model. During operation, the system captures and
matches a single face in real-time against the stored dataset. Upon recognition, attendance records are

IJFMR240323705 Volume 6, Issue 3, May-June 2024 4

 International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: editor@ijfmr.com

updated instantly in the Firebase database. This approach minimizes manual intervention and ensures
efficient attendance tracking.

Multiple Instance Attendance:

For environments requiring simultaneous recognition of multiple faces (e.g., classrooms or conferences),
advanced computer vision and machine learning techniques are applied. Users register their faces through
images to train a robust model capable of identifying multiple individuals in real-time. The system
continuously scans the environment, updates attendance records promptly in Firebase as faces are
recognized, and ensures accurate attendance logging across diverse settings.
Following images are resultant with different angles:

Testing and Results:

The system's testing phase involved evaluating its performance with faces in various positions and
orientations. Results demonstrate its capability to accurately detect and recognize individuals under
different conditions, validating its robustness. Real-time updates to attendance records in Firebase Real-
time Database reflect the system's effectiveness in maintaining up-to-date attendance information across
devices connected to the internet.

Conclusion:
a. Findings: Integration of technologies, Efficiency in recognition, Data handling, User interface, Single
interface recognition attendance, Multiple interface recognition attendance, Error handling of past
attendance system, provide accuracy and time-saving in recording attendance.
b. Suggestions: Enhanced Accuracy in Face Recognition Systems Using Deep Learning Techniques,
Scalability and Performance Optimization of Real-Time Face Recognition Systems, User Experience

IJFMR240323705 Volume 6, Issue 3, May-June 2024 5

 International Journal for Multidisciplinary Research (IJFMR)
E-ISSN: 2582-2160 ● Website: www.ijfmr.com ● Email: editor@ijfmr.com

Enhancements in Biometric Attendance Systems, Data Privacy and Security in Cloud-Based Biometric
Systems, Real-Time Database Management in Face Recognition Attendance Systems Using Firebase.

References:
1. L. D. H. a. A. B. L. A. J. Goldstein, "Identification of Human Faces," Proc. IEEE Conference on
Computer Vision and Pattern Recognition, vol. 59, pp. 758-760, May 1971.
2. M. A. Fischler and R. A. Elschlager, "The Representation and Matching of Pictorial Structures," in
IEEE Transaction on Computer, 1973.
3. S. S. R. Abibi, "Simulating Evaluation: connectionist metaphors for studying human cognitive
behaviour," in Proceedings TENCON 2000, vol. 1, pp. 167-173, 2000.
4. J. S. J. S. L. M. P. a. S. S. L. A. Y. Cui, "Automated Pattern Recognition and Defect Inspection
System," in in proc. 5th International Conference on Computer Vision and Graphical Image, May
1992.
5. M. K. D. A. S. a. D. R. P. Nirmalya Kar, "Study of Implementing Automated Attendance System,"
International Journal of Computer and Communication Engineering, vol. 1, p. 2, 2, July 2012

URLs:
1. Mortaza’s workshop and Robotics: https://youtu.be/iBomaK2ARyI?si=H_dsqWkjq dmI5uyd
2. Ws cube Tech: https://youtu.be/JZZr0PjZsIk?si=9DqHVYOV_ oB4DZ5U
3. Pysources: https://youtu.be/5yPeKQzCPdI?si=kxaxMDkH Y_yenE1P
4. Firebase Real-time Database: https://console.firebase.google.com/u/0/
5. Design of encoding: https://www.canva.com/

IJFMR240323705 Volume 6, Issue 3, May-June 2024 6