Employee attendance system using face recognition and GPS using local binary pattern histogram

Abstract

Tracking employee attendance is an integral part of running a company in an organized and economical manner. Conventional approaches such as manual sign-ins and RFID cards or fingerprint scanning have shown important weaknesses, especially with regard to proxy attendance (buddy punching). We chose the LBPH algorithm since it has a higher flexibility against changes of light, which means that we can use it in many situations like indoor or outdoor cases. The system performances for various conditions were also noteworthy, achieving 96.4% recognition accuracy with FAR = 0.05 %, FRR = 1 % in normal lighting conditions and maintaining a 94.1 % near-accurate performance under low-light environmental settings whilst sustaining the performance at 90.6 % in outdoor environments, which resulted in detection time of approximately between 1.3–2.3 seconds respectively. For further peace of mind, the system incorporated GPS tracking to provide location verification with a 90% to 94% accuracy rate—logging attendance only when students were present in a designated area. This integrated system is especially useful in contemporary hybrid workplaces, as it minimizes attendance fraud and enhances operational efficiency. Although the system is capable of functionally robust performance under normal conditions, tests point to possible scalability and performance improvements in extreme lighting conditions and outdoor applications, thus establishing future development paths for environmental adaptation.

Keywords

• face recognition
• gps tracking
• employee attendance system
• local binary pattern histogram (lbph)