<![CDATA[Efficiency, identity verification, and the possibility of duplicate attendance records are still issues with traditional attendance systems. The goal of this project is to create a Face ID-based attendance system that incorporates liveness detection, facial landmark detection, and face recognition into an online application. YuNet was used to recognize faces, MediaPipe Face Mesh was used to extract 478 facial landmark points, and SFace was used to create face embeddings and carry out cosine similarity matching. The Eye Aspect Ratio (EAR) method was used to detect eye blinks in order to accomplish liveness detection. The registration procedure successfully processed all 15 face frames and produced a 128-dimensional embedding, according to implementation results. While other tests produced scores as high as 0.945, face recognition testing effectively identified registered users with a similarity score of 0.921. Additionally, the system recorded attendance as "PRESENT," identified liveness as "REAL," and stopped duplicate entries by designating a "DUPLICATE" status. Testing with unidentified faces produced a "UNKNOWN" classification, which meant that attendance was not noted. In addition, the system offers a dashboard and reporting tools with data filtering and export features, sends email notifications, and stores data in a database. These findings show how the created system successfully combines facial recognition, liveness verification, attendance confirmation, notifications, and reporting into a single web-based platform]]>
