SISTEM PENGENALAN WAJAH MENGGUNAKAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK (CNN)

Abstract

The use of passwords is common in digital security and verification systems, but it is still vulnerable to manipulation, hacking, and theft. Alternative methods that can be done such as facial recognition have begun to be used, because facial features are difficult to fake, stable throughout life, and unique to each individual. Facial recognition can be done using various methods, one of which is using Convolutional Neural Network (CNN). CNN is a modern method based on deep learning that offers higher accuracy and processing speed, but requires greater computing resources. This study uses the CNN method to recognize faces in digital security and verification systems, with a dataset from the Yale Face Database, which contains 560 black and white facial images from 28 different subjects with varying formal expressions and lighting conditions, divided into 80% training data (448 images) and 20% testing data (112 images). The results of the study using the CNN method showed that its performance was quite good in terms of computational time efficiency for facial recognition. From the test results, it can be seen that CNN has a level of facial recognition accuracy, which is 98.6607%, in addition, the computation time for the CNN algorithm is quite fast, with a speed of 0.0030 seconds per image. This shows the superiority of CNN in capturing complex features of facial images and efficiency in data processing. However, it should be noted that the CNN algorithm requires quite a lot of computing resources.