TrueFace: a Dataset for the Detection of Synthetic Face Images from Social Networks

With today's technologies, the possibility to generate highly realistic visual fakes is within everyone's reach, leading to major threats in terms of misinformation and data trustworthiness. This holds in particular for synthetically generated faces, which are able to deceive even the most experienced observers, and can be exploited to create fake digital identities with synthetic facial attributes, to be used on social networks and online services. In response to this threat, researchers have employed artificial intelligence to detect synthetic images by analysing patterns and artifacts introduced by the generative models. However, most online images are subject to repeated sharing operations by social media platforms. Said platforms process uploaded images by applying operations (like compression) that progressively degrade those useful forensic traces, compromising the effectiveness of the developed detectors. To solve the synthetic-vs-real problem “in the wild”, more realistic image databases are needed to train specialised detectors. In this work, we present TrueFace, a first dataset of social-media-processed real and synthetic faces, obtained by the successful StyleGAN generative models, and shared on Facebook, Twitter and Telegram. The dataset is used to validate a ResNet-based image classification model addressing the discrimination of synthetic-vs-real faces in both presocial and post-social scenarios. The results demonstrate that even detectors with extremely high performance on non-shared images struggle to retain their accuracy on images from social media, while fine-tuning with shared images strongly mitigates such performance issues.

1. Introduction

Visual contents are the primary information vehicle on social media platforms, with over 3.2 billion images and 720,000 hours of videos uploaded every day [33]. The reason for their popularity is actually rooted in the very structure of the human brain, which is extremely fast and efficient at processing visual information, as opposed to, e.g., textual content. Visual media grab more attention from users, engage them at a higher level, and significantly increase the likelihood of sharing. At the same time, social networks are responsible for the viral diffusion of information worldwide, and play a key role in the digital life of individuals and societies. It is therefore unsurprising that there has been a long-standing interest by malicious users and organizations in manipulating visual contents and using them to spread unreliable information and fake news [38]. Figure 1.

The trueface dataset. The pre-social collection includes 70k real faces and 80k gan-generated images, half from stylegan [13] and half from stylegan2 [14]. The post-social collection includes the shared version of a portion of the pre-social part, where images have been uploaded and downloaded from facebook, telegram, and twitter, for a total of 60k images. The dataset is available at https://bit.Ly/3baeh75.