Real-Time CNN-Based Face Recognition Platform for Scalable Identity Management and Alert Transmission
1A.Harsha Vardhan Reddy, 2A.Saiprashanth, 3S.Praneeth Reddy, 4Ch. Raja
1,2,3 Department of ECE, Mahatma Gandhi Institute of Technology (A), Gandipet, Hyderabad, Telangana, India. Email: anthireddyh@gmail.com
4 Associate Professor, Department of ECE, Mahatma Gandhi Institute of Technology (A), Gandipet, Hyderabad, Telangana, India. Email: chraja@mgit.ac.in
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - This paper presents a comprehensive, cloud-integrated face recognition-based attendance system that combines machine learning, web technologies, and scalable backend architecture. The backend server, built with Node.js and MongoDB, securely stores user data, embeddings, and supports authentication via JWT and RBAC. The core recognition pipeline uses MTCNN for face detection, FaceNet for embedding generation, and SVM for classification. Recognized faces trigger SMS alerts via Twilio, with optional email notifications. A React frontend allows image uploads, result viewing, and user management, while a Flask layer bridges UI and ML logic. The system is deployed on AWS using Docker, ECS, and MongoDB Atlas, with CloudFront enhancing frontend delivery. CI/CD pipelines automate testing and deployment, while CloudWatch and Prometheus ensure robust monitoring. Auto-retraining of the SVM model accommodates new users, and comprehensive backup strategies with multi-zone redundancy ensure disaster recovery. Security enhancements such as hashed passwords, rate limiting, CORS policies, and encrypted cloud storage further strengthen system resilience. Designed for scalability, the platform auto-scales backend containers, caches frequent queries using Redis, and employs load balancing for high availability. This modular, end-to-end solution provides a robust framework for real-time, secure, and scalable facial recognition applications across educational or enterprise environments.
Key Words: Automatic Attendance System, Convolutional Neural Networks (CNNs), Facial Recognition, MTCNN, Smart Attendance Monitoring, FaceNet.