Convolutional Neural Network (CNN) on Realization of Facial Recognition Devices using Multi Embedded Computers

Abstract

Facial recognition is a field that is still being researched and developed. However, artificial intelligence, particularly the artificial neural network (JST) Convolutional Neural Network, is often used in facial recognition systems. The Convolutional Neural Network method is a supervised learning method (laryngeal supervision) commonly used, so it is appropriate for facial recognition applications. In this application, image processing of input imagery before the image is processed in artificial neural networks using the Convolutional Neural Network method.The study used the Convolutional Neural Network method on facial recognition. The way to be used in the facial recognition system to be obtained, disadvantages, and advantages of the method used. (face recognition). This research can be analyzed, namely accuracy in each learning rate and get results. The accuracy factor is derived from how precisely the person's name is recognized from the face image.In facial recognition systems with the Convolutional Neural Network (CNN), it has analysis results that refer to increased accuracy. This is due to the Convolutional Neural Network structure itself, an optimization of the Backpropagation method. With the specific role of the Convolutional Neural Network method that has been specified to solve the problem of image recognition. For the result testing; for the Front Side detection, the best accuracy is 85.60% on the learning rate (α) 0.01 and Epoch 30 for object 3 (K), for the Left Side detection, the best accuracy is 92.96% on the learning rate (α) 0.01 and Epoch 20 for object 3 (K), and for the Right Side detection the best accuracy is 95.36% on the learning rate (α) 0.01 and Epoch 25 for object 3 (K).