<![CDATA[Student Attendance Systems that still rely on manual or semi-manual methods are often prone to recording errors and misuse, which can disrupt the academic evaluation process. Facial recognition technology can offer a solution by enabling the unique identification of individuals based on facial features and allowing automatic, real-time attendance recording. This study aims to develop a facial recognition attendance system using Google ML Kit and FaceNet in a mobile application. Testing was conducted under various conditions, including different distances, lighting, and the use of accessories, to evaluate the system's reliability in real-world scenarios. The results show 100% accuracy at distances of 50 cm, 100 cm, and 150 cm, although recognition time slowed from 1.328 seconds at 50 cm to 1.963 seconds at 150 cm. Accuracy decreased in low-light conditions, and the simultaneous use of accessories such as hats and glasses reduced accuracy to 78.75%. Additionally, the system exhibited a False Acceptance Rate (FAR) of 10% when tested with faces outside the database. Overall, the developed facial recognition system demonstrates high accuracy under ideal conditions but still requires optimization for varying environmental conditions.]]>
Copyrights © 2025
