<![CDATA[Identifying a person's self-identity can be done by recognizing facial images, where faces can often represent a person's identity. Facial identification with technology can benefit the effectiveness efficiency and accuracy of data. This identification process can be used with the help of algorithms that will check digital images with the necessary detection results. One algorithm that can be applied in classifying and detecting gender through facial image algorithms is Convolutional Neural Networks. Convolutional Neural Network algorithms have various architectures that have advantages in each architecture. This study compared the process of identifying a person's face to obtain information in the form of gender. The models compared in this study are the LeNet model and the VGG16 model. The identification and detection process was carried out using 800 photos for data training with gender labeling data and 240 photos for testing data. A comparison of these two models is necessary to get the best final model result. The final results obtained from this study the best accuracy of both architectures was obtained in the VGG16 architecture which reached an average accuracy of 100 in several epochs compared to the VGG16 architecture at 0.925 in the 46th epoch. This is due to a Rectified Linear Unit (ReLU) on the VGG16 architecture which can minimize errors and saturation.]]>
Copyrights © 2023
