Markit: An Automatic Facial Recognition-Based Attendance Management System

Abstract

Tracking classroom attendance is often tedious and error-prone, especially regarding proxy attendance. Techniques commonly used to take roll calls and use physical registration systems are disruptive to classroom processes and unfortunately not always as reliable. MarkIt provides an online system that takes roll calls by using a basic web camera and is totally web based, meaning no sophisticated hardware is needed to utilize it. MarkIt employs YOLOv8 to detect faces in real-time and generates facial embedding vectors via face-api.js, which are matched to individual student profiles through usage of Euclidean distance and permits for reliable identification. MarkIt is developed on a React/TypeScript stack, with Supabase providing authentication/storage support, computations being done locally utilizing WebGL and WebAssembly; providing low latencies for processing while preserving the highest quality of privacy because no sensitive info ever leaves the client. MarkIt offers an expedited, accurate and secured process to track and maintain student attendance.