Face swapping combines one face's identity with another face's non-appearance attributes (expression, head pose, lighting) to generate a synthetic face. This technology is rapidly improving, but falls flat when reconstructing some attributes, particularly gaze. Image-based loss metrics that consider the full face do not effectively capture the perceptually important, yet spatially small, eye regions. Improving gaze in face swaps can improve naturalness and realism, benefiting applications in entertainment, human computer interaction, and more. Improved gaze will also directly improve Deepfake detection efforts, serving as ideal training data for classifiers that rely on gaze for classification. We propose a novel loss function that leverages gaze prediction to inform the face swap model during training and compare against existing methods. We find all methods to significantly benefit gaze in resulting face swaps.

Due to their convenience and high accuracy, face recognition systems are widely employed in governmental and personal security applications to automatically recognise individuals. Despite recent advances, face recognition systems have shown to be particularly vulnerable to identity attacks (i.e., digital manipulations and attack presentations). Identity attacks pose a big security threat as they can be used to gain unauthorised access and spread misinformation. In this context, most algorithms for detecting identity attacks generalise poorly to attack types that are unknown at training time. To tackle this problem, we introduce a differential anomaly detection framework in which deep face embeddings are first extracted from pairs of images (i.e., reference and probe) and then combined for identity attack detection. The experimental evaluation conducted over several databases shows a high generalisation capability of the proposed method for detecting unknown attacks in both the digital and physical domains.

The website is eye-catching for its simplicity. Against a white backdrop, a giant blue button invites visitors to upload a picture of a face. Below the button, four AI-generated faces allow you to test the service. Above it, the tag line boldly proclaims the purpose: turn anyone into a porn star by using deepfake technology to swap the person's face into an adult video. All it requires is the picture and the push of a button.

Buzzy face-swapping video app Reface is expanding its reality-shifting potential beyond selfies by letting users upload more of their own content for its AI to bring to life. Users of its iOS and Android apps still can't upload their own user generated video but the latest feature -- which it calls Swap Animation -- lets them upload images of humanoid stuff (monuments, memes, fine art portraits, or -- indeed -- photos of other people) which they want animated, choosing from a selection of in-app song snippets and poems for the AI-incarnate version to appear to speak/sing etc. Reface's freemium app has, thus far, taken a tightly curated approach to the content users can animate, only letting you face swap a selfie into a pre-set selection of movie and music video snippets (plus memes, GIFs, red carpet celeb shots, salon hair-dos and more). But the new feature -- which similarly relies on GAN (generative adversarial network) algorithms to work its reality-bending effects -- expands the expressive potential of the app by letting users supply their own source material to face swap/animate. Some rival apps do already offer this kind of functionality -- so there's an element of Reface catching up to apps like Avatarify, Wombo and Deep Nostalgia. But it's also going further as users can also swap their own face into their chosen source content.

In a few short years, neural-network-powered automated face swaps have gone from being mildly convincing to eerily believable. But through new research from Disney, neural face-swapping is poised to become a legitimate and high-quality tool for visual effects studios working on Hollywood blockbusters. One of the bigger challenges of creating deepfake videos, as they've come to be known, is creating a vast database of facial images of a person--thousands of different expressions and poses--that can be swapped into a target video. The larger the database and the higher the quality of the images, the better the face swaps will turn out. But the images (which are more often than not headshots of famous people) are usually pulled from sources with limited resolution.

Chinese social media upstart, TikTok and its counterpart Douyin are turning to technology commonly used for creating deepfakes to power a yet-to-be-released feature. According to a report from TechCrunch, ByteDance, which owns TikTok and China-based Douyin, has been developing a feature that allows users to create videos in which their face is superimposed onto someone else's. The feature, which mirrors other deepfake technology used to doctor videos of politicians and public figures, is being referred to as'Face Swap' within TikTok's own code according to TechCrunch and has not yet been released to users. The face swapping feature, while similar to those long-used by other social media platforms like Snapchat, differs in its ability to realistically superimpose faces on videos according to TechCrunch. 'Face Swap' reportedly works by taking a biometric scan of a users' face from multiple angles - similar to the process of setting up a facial recognition app like Apple's Face ID - and then lets users choose videos that they want to insert their face onto.

When Peter Cushing turned to face the camera in Rogue One, Star Wars fans were as excited as they were confused. After all, the actor had died more than 20 years earlier, and yet, there was no mistaking him. For a major Hollywood movie, this is a clever trick. But not everyone is trying to entertain us, and you don't need a million-dollar budget to deceive. "You take the face of one person and put it on the body of another," said Jeff Smith, associate director at the National Center for Media Forensics at the University of Colorado Denver.

The ability to take one person's face or expression and superimpose it onto a video of another person has recently become possible. In particular, pornographic videos called "deepfakes" have emerged on websites such as Reddit and 4Chan showing famous individuals' faces superimposed onto the bodies of actors. This phenomenon has significant implications. At the very least, it has the potential to undermine the reputation of people who are victims of this kind of forgery. It poses problems for biometric ID systems.

The team, led by Andreas Rossler at the Technical University of Munich, developed machine learning that is able to automatically detect when videos are face swapped. They trained the algorithm using a large set of face swaps that they made themselves, creating the largest database of these kind of images available. They then trained the algorithm, called XceptionNet, to detect the face swaps. XceptionNet clearly outperforms its rival techniques in detecting this kind of fake video, but it also actually improves the quality of the forgeries. Rossler's team can use the biggest hallmarks of a face swap to make the manipulation more seamless.

In December 2017, a user named "DeepFakes" posted realistic looking explicit videos of famous celebrities on Reddit. He generated these fake videos using deep learning, the latest in AI, to insert celebrities' faces into adult movies. In the following weeks, the internet exploded with articles about the dangers of face swapping technology: harassing innocents, propagating fake news, and hurting the credibility of video evidence forever. It's true that bad actors will use this technology for harm; but given that the genie is out of the bottle, shouldn't we pause to consider what else DeepFakes could be used for? In this post, I explore the capabilities of this tech, describe how it works, and discuss potential applications.