Pengujian Algoritma MTCNN (Multi-task Cascaded Convolutional Neural Network) untuk Sistem Pengenalan Wajah

Abstract

Measurement of facial similarity or checking similarity is done using features. The algorithm for describing the most up-to-date and best face features for generating features is Deep Convolutional Neural Network (DCNNs). Based on this, this study uses MTCNN (Multi-task Cascaded Convolutional Neural Network) as one variation of the DCNN method. In this research, we built a research system to test results with javascript. Given the many needs that are based on mobile or can be run on a smartphone. One of them is to support the absent feature that is used in a mobile manner such as the reporting system of sales and marketing performance or members of the police personnel who normally work on a mobile basis. From the results of the tests carried out automatically using several variation models testing the image of the Aberdeen dataset as many as 60 images from 30 different people used in the face recognition research system using MTCNN with influencing image parameters such as lighting variations, object position variations, then the position taken and expression face on the object image, the research system managed to do face recognition by 100%. Thus, true positive values are equal to the amount of data tested and zero negative true values.

Value of Virtual Reality Technology in Image Inspection and 3D Geometric Modeling

“…In order to verify the effectiveness of the cascaded convolutional neural network model and training method designed in this paper, using the same training and test data, five target detection methods based on convolutional neural networks were compared: SSD300 [31], YoLoV2 [32], FRCNN [33], RetinaNet [34] and MTCNN algorithm [35]. SSD300 uses VGG16 [36] as the backbone network, and YoLov2, FRCNN and RetinaNet use ResNet50 as the backbone network.…”