Smart AI-based CCTV attendance ledger using face recognition and liveness validation

Abstract

Taking attendance is a small task that happens regularly, but it can take up valuable class time. Paper registers and manual roll calls also leave room for missed entries, typing mistakes, and proxy attendance. This paper describes a Smart AI-Based CCTV Attendance Ledger designed to record attendance from CCTV/IP-camera streams or USB webcams. The project repository documents a modular application using Next.js, React, TypeScript, FastAPI, Python, OpenCV, DeepFace, and PostgreSQL through Supabase, with SQLite available for local development. The proposed workflow combines face detection, identity matching, liveness or anti-spoofing checks, and persistent attendance records. For the system to be useful in practice, it also needs to handle poor-quality images, repeated detections, uncertain matches, database errors, access permissions, and the privacy risks that come with biometric data. This paper explains the problem, related work, system design, operational workflow, evaluation strategy, risks, and future improvements. This paper does not report accuracy or speed figures because no reproducible test results were available when it was prepared.