GAN to Produce New Faces and Detection Expression

Abstract

AbstractIn a world where data is the most significant item. Image has now set off to be the most valuable data. Along with the evolving technologies, image is being bifurcated and used in the field of machine learning for various operations. From the recent past, image is bifurcated into segments and could be seen to be used everywhere, whether it is predicting market patterns or imitating the real world in a form of virtual grid. Segmentation of images is the most critical part of developing a machine learning model. Better is the training of the model, better will be the results, which may result in a successful machine learning model. A recent innovative proceeding has been introduced related to image segmentation, generative adversarial network (GAN), which will surely bring huge achievements in this field. In this study, the function of GAN in producing new faces and subsequently detecting facial expressions is elaborated. This study can be divided into two phases, the first is to generate faces using GAN, and the second phase is to detect expressions of the faces generated. For the first phase, the aim is to train the generator model with the existing face dataset and after the successful training, the generator will be able to produce new faces. In the second phase, a model is prepared which can extract the facial features and detect expressions based on these features.KeywordsMachine learningGenerative adversarial networks (GAN)Convolutional neural network (CNN)Face generationFacial recognitionExpression detection