FRAS: AI Classroom Attendance

Summary

FRAS (Facial Recognition Attendance System) is an AI-based system developed to automate attendance tracking in academic classrooms using computer vision. Designed as a year-long senior capstone project during my Computer Science program, FRAS aims to solve the inefficiencies of traditional attendance-taking methods, especially in classes with high enrollment.

Team Members and Roles

• Shawki Izzat, Team Leader: Proposed the initial concept, acted as the liaison between faculty and team members, designed the backend database, led UI/UX efforts, and created the final project demo video.

• Hassan Al Ali, AI Expert: Specialized in Python and facial recognition algorithms. Selected and implemented the optimal facial detection model, and managed backend integration.

• Iyas Naser, Desktop Developer: Developed the professor-facing desktop application in Java and assisted in integrating software and hardware components.

• Mohamad Saleh, App Architect: Developed frontend and backend for mobile application, and collaborated on refining UI elements throughout.

Inspiration

As engineering students, we witnessed the cumbersome process of manually taking attendance, particularly in classes with high class strength. Traditional methods—like calling out names or scanning barcodes—were not only time-consuming but also easy to exploit and failed to track late arrivals. This sparked our interest in using facial recognition to create a seamless, secure, and background attendance process.

Objectives

• Automate attendance-taking without disrupting class time

• Ensure high facial recognition accuracy

• Respect privacy—no image storage

• Design a simple, intuitive interface for students and professors

Challenges

• Selecting a reliable facial recognition algorithm that functions accurately in varied classroom conditions

• Balancing privacy with functionality

• Adapting to COVID-19 restrictions in the final stages

Design and Development Journey

Fall 2019: Research and Prototyping

• Began with brainstorming and researching existing solutions

• Evaluated hardware requirements like panoramic cameras ideal for facial detection range

• Assessed database and cloud options with scalability and privacy in mind (Firebase and Azure)

• Ran simulated classroom tests with volunteer students to assess facial recognition accuracy

• Achieved a 93% accuracy rate using a panoramic camera and eigenvalue-based face recognition, confirming the feasibility of our core algorithm

We made key architectural decisions during this phase, including:

• A modular system separating student and instructor workflows

• An image-free recognition process using mathematical face vectors

• Time-stamped image intervals for tracking attendance across an entire lecture

Spring 2020: Implementation

• Built a student mobile application (Android Studio, Java) where users registered with a photo that was converted into eigenvalues and then discarded

• Designed a professor desktop application (Java) for classroom-based attendance, allowing faculty to start and stop attendance sessions and view historical data

• Developed time-interval-based logic to categorize attendance (present, late, absent) using snapshots at defined periods across a class session

• Integrated all components via Firebase and Microsoft Azure, ensuring real-time syncing and secure data storage

Each classroom session was divided into four time intervals to track presence dynamically. This logic helped us account for students arriving late or stepping out briefly while still ensuring fairness and accuracy.

COVID-19 Adaptation

In March 2020, lockdowns forced us to work remotely from different countries. Without access to physical classrooms, we pivoted by:

• Simulating a classroom through video recordings

• Testing the FRAS system on multi-person videos at various simulated row depths

Despite the shift, our system yielded a 97% accuracy rate during the final remote demo.

Reception

The FRAS project was well received by our mentors and peers, especially given the difficult context of the pandemic. It was considered one of the most advanced projects in our cohort.

Reflection

FRAS was the most technically challenging and rewarding project of my academic career. It taught me about leadership, cross-functional collaboration, and the importance of flexibility in the face of unforeseen challenges. What began as a concept born out of classroom frustration evolved into a functioning prototype with real-world implications—and that journey remains one of the proudest accomplishments of my time as a student engineer.