<![CDATA[The attendance system at SMK Taruna Terpadu 1 with nine majors is still done manually. With a total of about 5,000 students, If attendance is recorded manually, many of these statistics are cumulative, making them difficult to organize and find when needed. Digitization of attendance recording is expected, one of which is the biometric method. Biometrics, the technology that digitally recognizes organic characteristics, can potentially update maps and other identifiers. Biometrics themselves come in physical form, such as faces, irises, fingerprints, and handprints. However, at some point during the COVID-19 pandemic, contact fingerprinting is unavailable and many of the challenges facing facial recognition, starting with skin color, using mask and identical twins. suggest ways to avoid contact. Fingerprint biometrics are an attractive option for more accurate, reliable, and secure contactless human identification technology, but identifying palm features from past images is also an attractive option. I am tasked with inputting some of the palm functions. and lighting fixtures. In this article, the authors propose to apply MobileNeV2's use of augmented facts, ROI detection, and pre-trained convolutional neural community (CNN) models. After testing with the dataset that the author got from SMK Taruna Terpadu 1 by performing data augmentation, ROI detection and identification with the pretrained MobileNetV2 model, it turns out to get the best accuracy results up to 99.98%.]]>
