Abstract

This study presents Nexus-AI, a smart attendance system that integrates facial recognition and autonomous agents to automate attendance tracking and streamline academic communication. Built using Python, Flask, OpenCV, and GPT-3.5, the system enables real-time check-ins, personalized chatbot queries, and dynamic report generation. Testing was conducted with 15 participants in varied conditions to evaluate accuracy, speed, and user interaction. Results showed 80–100\% recognition accuracy in normal lighting and frontal views, but significant performance drops (0–20\%) under low-light or occluded conditions. Despite these challenges, the system consistently maintained a 0\% false acceptance rate and operated with sub-second response times across most components. Autonomous agents responded to natural and rephrased queries, automating tasks like absence alerts, prediction, and insights with high reliability. While limitations remain—such as registration time, lighting sensitivity, and single-camera dependency—the system effectively reduced manual workload and improved transparency in attendance tracking. Ethical safeguards, including local data storage and informed consent, ensured responsible AI use. Nexus-AI contributes to the growing field of intelligent classroom systems by demonstrating how multi-agent AI can enhance both efficiency and engagement in real educational settings.

Advisor

As'ad, Asa’d

Department

Computer Science

Publication Date

2025

Degree Granted

Bachelor of Arts

Document Type

Senior Independent Study Thesis

Download

Share

COinS
 

© Copyright 2025 Siham Boumalak