MUMBAI, India, June 26 -- Intellectual Property India has published a patent application (202641052482 A) filed by Hindusthan College Of Engineering And Technology on April 24, 2026, for Automated Student Attendence Monitoring.
Inventors include Karthik S; Haridass S; Arul S; Aswin M; and Kalanithi J.
The application for the patent was published on June 19, 2026, under issue no. 25/2026.
Abstract: ABSTRACT This project presents an Al-based Attendance Management System that automates student attendance using deep learning and real-time processing techniques. The system captures live video through a high-resolution camera, where each frame is pre-processed and analyzed using the MTCNN algorithm for accurate face detection and alignment. The detected faces are then passed to the FaceNet model, which generates a unique 512-dimensional facial embedding representing individual features as numerical vectors. During the enrollment phase, multiple facial samples are collected and averaged to create a master embedding for each student, ensuring higher recognition accuracy. In the identification phase, live embeddings are compared with stored embeddings using cosine similarity and Euclidean distance to determine matches. To improve reliability and reduce false positives, a streak-based verification mechanism is implemented, requiring consistent recognition across multiple frames. Once verified, the system automatically records attendance by sending asynchronous requests to a Node.js and Express backend, where data is stored in a MySQL database with timestamps and session details. The frontend, developed using React, provides a user-friendly interface with real-time visual feedback. Additionally, the system identifies absentees by comparing session records and sends notifications via the Twilio SMS API to administrators or guardians. Security is maintained through non-reversible storage of facial embeddings, while performance is optimized using caching and vectorized computations
Disclaimer: Curated by HT Syndication.
