Implementation of Fingerprint Recognition Using Convolutional Neural Network and RFID Authentication Protocol on Attendance Machine
Maredi Aritonang, Irwan Doni Hutahaean, Hasudungan Sipayung, Indra Hartarto Tambunan · 2020
The attendance machine is a machine that can record a person's attendance data at an institution or office that applies it. The current attendance system is considered to be less effective because it still implements a manual system that has weaknesses in its use, such as many paper usage and opening gaps to falsify data. One effort to solve this problem is to use fingerprint and RFID attendance machine. In this research, the fingerprint grouping is performed so the counterfeit presence data can be minimized due to the unique and identical fingerprint pattern. The process of grouping fingerprint images requires an approach that uses the convolutional Neural Network (CNN) algorithm because of the difficulty of distinguishing fingerprint patterns. Based on the results of the implementation, it has managed to obtain an accuracy rate of 95.64%, validation accuracy of 98.76%, and a loss of 0.001% against 2 image classes. The attendance machine also uses RFID technology as an alternative if the fingerprint system is experiencing interference. In this research, the RFID authentication process uses the RC4 (Rivest Code4) cryptographic algorithm to encrypt the Unique Identifier (UID) of the student card. Attendance machine built using Raspberry Pi 3 microcontroller integrated with the GT-521F52 type fingerprint sensor and RFID RC522. The system saves the recorded data in the database and connects to server so that it can be accessed by the user via the website.