CD Skripsi
Implementasi Face Recognition Menggunakan Cnn Pada Buku Tamu Elektronik Di Badan Pusat Statistik Provinsi Riau
Traditional guestbook systems face challenges such as lost or damaged paper, long waiting times, and inconsistent data entries. While digital guestbooks have addressed some of these issues, repeated data entry by frequent visitors remains a problem. By implementing face recognition, visitors only need to fill in their identity during the first visit, with subsequent visits automatically recognized through facial features stored in the database. The implementation of face recognition using CNN in the electronic guestbook at the Central Bureau of Statistics (BPS) of Riau aims to improve the efficiency and accuracy of recording visitor identities. The biometric face recognition technology uses CNN, comprising face detection, face embedding, and face identification stages, chosen for its high accuracy in facial recognition. The CNN MediaPipe BlazeFace model is used for face detection, while facial feature extraction is performed by the CNN HSE FaceRes model. Face identification is conducted by measuring the distance between facial features using Square Euclidean Distance. Testing shows the system's accuracy reaches 100% for primary test data and 87% for secondary test data. Additionally, the time taken for face recognition is more efficient compared to manual entry, with an average time difference of 54.32 seconds and an efficiency of 74.48%. This implementation is expected to facilitate visitors, enhance administrative efficiency, and ensure the accuracy of visitor data at BPS Riau Province.
Keywords: Face Recognition, Electronic Guestbook, Convolutional Neural Network (CNN)
Tidak tersedia versi lain