Enhancing Face-Based Detection of Attendance Validation System

Abstract

Attendance systems are essential for organizational compliance but often rely on non-visual mechanisms, such as QR codes or location verification. These conventional methods remain vulnerable to invalid attendance records, such as proxy attendance. This study investigates the implementation of a face-based detection approach for a web-based attendance validation system. It integrates facial image capture with QR code interaction, location verification, and time-based logging. The system utilizes lightweight 1:1 facial verification to confirm the physical presence of users during attendance events. Each attendance attempt records a facial image, timestamp, geographic coordinates, and attendance type to ensure that validation occurs only when all constraints are satisfied. Experimental evaluation involves 100 attendance attempts under three validation scenarios: normal conditions, facial occlusion cases, and invalid location attempts. Under normal conditions, the system correctly validates 48 out of 50 legitimate attendance attempts, achieving 96.0% accuracy, with two false rejections caused by temporary lighting limitations during image capture. For invalid scenarios, the system successfully rejects all 30 attempts involving facial occlusions such as masks, hats, and glasses, achieving 100% rejection accuracy. Similarly, all 20 attempts conducted outside the authorized geographic radius are blocked, also achieving a 100% rejection rate. Overall, the integrated validation framework achieves a system-level accuracy of 98.0%. The results demonstrate that face-based presence detection can provide a reliable mechanism for attendance validation.