Pengenalan Wajah Pegawai Kantor Dengan Menggunakan Metode Convolutional Neural Network Berbasis Android

Abstract

Face is one of the elements used to identify identity between humans. The purpose of making this thesis is as a basic basis for developing an attendance system and making artificial intelligence that can identify humans through their faces. How to do data processing, the data taken comes from a video of office employees which lasts approximately 10 seconds. To make a program that can recognize the faces of office employees, the Convolutional Neural Network (CNN) method is used which will be trained to be able to distinguish each unique feature on the face to distinguish and recognize humans specifically. In performing facial recognition, office employees can provide input in the form of facial photos of office employees who have been trained and use the camera on a smartphone to perform face recognition directly. The faces of office employees used as targets for this CNN training came from Pt Eternal Indonesia, Faculty of Information Technology, Tarumanagara University, and Kekar ​​Clinic. The output of the application is the accuracy of each photo of the office employee's face given. The results of the confusion matrix test show that the trained model has an accuracy of 80.39%, a precision of 80%, a recall of 80%, and an f1-score of 80%.