IDENTIFIKASI GARIS TELAPAK TANGAN DENGAN METODE MOBILENET CONVOLUTIONAL NEURAL NETWORK (CNN) UNTUK SISTEM PRESENSI SISWA

Abstract

The attendance system currently used at SMK Taruna Terpadu 1, which has nine majors and around 5.000 students, is done manually. This leads to many statistics being recorded by hand, making it difficult to keep track of and find when needed. To solve this problem, attendance recording is expected to be digitized using biometrics, which are computer-based recognition systems that recognize physical traits. Biometrics can come in the form of faces, irises, fingerprints, and handprints. However, contact fingerprinting will be unavailable during the COVID-19 pandemic, and facial recognition poses challenges such as skin color, mask use, and identical twins. Therefore, fingerprint biometrics and palm features are more attractive options for contactless human identification technology, as they are more accurate, reliable, and secure. This article proposes the use of MobileNeV2's augmented facts, ROI detection, and pre-trained convolutional neural network (CNN) models to identify palm features and lighting fixtures. The authors used data augmentation, ROI detection, and identification with the pre-trained MobileNetV2 model to test the dataset from SMK Taruna Terpadu 1 and achieved an accuracy result of up to 99.98%.