Face Recognition-Based Attendance System

Abstract

Traditional attendance recording methods used in educational institutions-such as manual sign-ins, RFID cards, or fingerprint scanners-are often time-consuming, error-prone, and vulnerable to impersonation or fraudulent practices. The growing shift toward digital and hybrid learning environments has further highlighted the limitations of these systems, particularly regarding accuracy, hygiene concerns, and scalability. To address these challenges, this research proposes a Face Recognition-Based Attendance System that leverages machine learning techniques to provide a fast, secure, and contactless method for attendance management.

The system integrates MTCNN for robust face detection and FaceNet for generating high-dimensional facial embeddings, which are then classified using a Support Vector Machine (SVM) model. A complementary LBPH-based pipeline is also implemented to enhance recognition performance under varying lighting conditions and lower-quality image inputs. Extensive preprocessing techniques—including image normalization, augmentation, and contrast enhancement—were applied to improve model accuracy and adaptability.

Experimental results demonstrate that the system performs effectively under diverse environmental conditions, including changes in lighting, background, and facial expressions. Both pipelines achieved strong real-time detection capability, with high accuracy in identifying registered individuals and reliably marking their attendance. The system reduces administrative workload, eliminates proxy attendance, and provides a hygienic alternative to traditional biometric systems.

This research contributes to both academic and practical domains by presenting a hybrid machine-learning approach that enhances recognition robustness while maintaining real-time performance. The findings highlight the potential of face recognition technologies as a scalable, efficient, and secure solution for modern attendance management systems, particularly in dynamic learning environments.