AUTOMATIC ATTENDANCE RECORDER

Abstract

A highly robust hybrid facial recognition–based automated attendance system is introduced, combining the strengths of classical computer vision with modern deep learning techniques to achieve fast and accurate identity recognition. For initial detection, Multi-Haar Cascade classifiers are utilized to locate faces efficiently, ensuring lightweight and real-time performance suitable for active classroom environments where lighting and student movement often vary. Once a face is detected, FaceNet is employed to generate 128-dimensional embeddings that capture unique and discriminative facial characteristics, making identification reliable even when there are changes in angle, expression, or distance.These rich embeddings are then processed through a Convolutional Neural Network (CNN) classifier trained to distinguish between individual identities with high precision. The combination of Haar-based detection and deep-learning-driven feature extraction contributes to a balanced system that offers both speed and accuracy. After successful recognition, attendance data is automatically generated in CSV format, making it easy for institutions to store, manage, and analyze records. The logs are seamlessly uploaded to a dedicated college-level administrative interface, allowing authorized personnel to review daily attendance, access summaries, and maintain long-term archives.