REAL-TIME WEB-BASED EMPLOYEE ATTENDANCE USING OPENCV AND LBPH FACE RECOGNITION

Abstract

This study developed a real-time web-based employee attendance system that applies face recognition using OpenCV and the Local Binary Pattern Histogram (LBPH) method. The research object was the employee attendance process at PT Samboeng Putra Margo Utomo, where conventional attendance created risks of recording errors, slow recapitulation, and proxy attendance. The system was developed using the Waterfall model through requirements analysis, system design, implementation, testing, and evaluation. The application integrates a web interface, employee and shift management, face registration, attendance check-in and check-out, leave requests, attendance reporting, a MySQL database, and an OpenCV-based face recognition module. Functional validation used black-box testing, while face recognition reliability was examined through repeated trials under variations in lighting, camera distance, slight face angle, and facial expression. All 40 functional test scenarios reported in the thesis were completed successfully. In the face recognition reliability test, 56 of 60 trials were successfully recognized, producing a recognition success rate of 93.33%. Normal lighting and close-distance conditions achieved 100% success, while dim lighting, medium distance, slight face rotation, and different expressions each achieved 90%. The results indicate that the developed system can automate attendance recording, integrate attendance administration, and reduce dependence on manual identity input. The findings also show that LBPH is suitable for a lightweight real-time attendance implementation, although performance remains sensitive to image acquisition conditions and should be strengthened with broader datasets and liveness detection in future work