Detection of Artificially Created Faces with Convolutional Networks

The automatic detection of artificially generated content presents one of the most current topics in the field of artificial intelligence. It may help disguise deceitful users, spread misinformation or help in the dissemination of false accusations and therefore it is may prove vital in terms of teaching and improving the information literacy at schools,public institutions or general public. The automatic recognition of various forms of toxicity and artificially created faces in online space can help teachers to teach an information literacy and a critical thinking.This paper focuses on the detection of artificially generated faces depicted on still images with the utilization various types of convolutional neural networks.The paper presents an experiment aimed at multitude of methods which share the same convolutional basis. The focus is the creation of a efficient tool for image classification, which for our purposes was conducted with the use of ResNet, DenseNet and VGG architectures.