Deepfake is an advanced technology that creates extremely realistic facial images and videos. This new technique operates under specific conditions and has a wide range of applications. For example, it can be used in the entertainment industry to create impressive visual effects or to insert actors into scenes convincingly. Similarly, in the film industry, deepfakes can help make movies by faithfully reproducing the appearance of actors who are not physically present. It is also useful for creating realistic digital avatars of people, which can be used in virtual environments, video games, or augmented reality applications. Recently, the emergence of new content generation models capable of creating impressively realistic images has been gaining momentum. Despite their advantages, they also cause significant issues when used maliciously, such as for identity theft, misinformation, and obscene depictions of well-known individuals. Therefore, it is crucial to implement effective methods to expose this generated content and thus reduce crime associated with deepfakes. This article presents a novel method for detecting fake content based on an in-depth analysis of the characteristics of eye irises. By applying a gradient map to the iris, it is possible to visualize the biological characteristics specific to eye irises, such as the round shape, identical reflections in the two irises of the same face, the size of the iris, etc. The gradient map highlights all the contours of the objects present in the iris; thus, the reflected light present in the corneas is represented by brighter pixels comparable to heat. We show that two irises of the same face are almost identical in shape, reflection, and size. Our experimental results on the Flickr-Faces-HQ (FFHQ) dataset and images obtained from StyleGAN2 demonstrate that our algorithm achieves a remarkable detection accuracy of 0.979 and 0.921 sensitivity. Furthermore, the method has a specificity of 0.937 and a precision of 0.960, thereby proving the effectiveness of the gradient map associated with the shape of the pupil in detecting Generative adversarial network (GAN) generated faces.

Currently, the phenomenon of deepfakes is becoming increasingly significant, as they enable the creation of extremely realistic images capable of deceiving anyone thanks to deep learning tools based on generative adversarial networks (GANs). These images are used as profile pictures on social media with the intent to sow discord and perpetrate scams on a global scale. In this study, we demonstrate that these images can be identified through various imperfections present in the synthesized eyes, such as the irregular shape of the pupil and the difference between the corneal reflections of the two eyes. These defects result from the absence of physical and physiological constraints in most GAN models. We develop a two-level architecture capable of detecting these fake images. This approach begins with an automatic segmentation method for the pupils to verify their shape, as real image pupils naturally have a regular shape, typically round. Next, for all images where the pupils are not regular, the entire image is analyzed to verify the reflections. This step involves passing the facial image through an architecture that extracts and compares the specular reflections of the corneas of the two eyes, assuming that the eyes of real people observing a light source should reflect the same thing. Our experiments with a large dataset of real images from the Flickr-FacesHQ and CelebA datasets, as well as fake images from StyleGAN2 and ProGAN, show the effectiveness of our method. Our experimental results on the Flickr-Faces-HQ (FFHQ) dataset and images generated by StyleGAN2 demonstrated that our algorithm achieved a remarkable detection accuracy of 0.968 and a sensitivity of 0.911. Additionally, the method had a specificity of 0.907 and a precision of 0.90 for this same dataset. And our experimental results on the CelebA dataset and images generated by ProGAN also demonstrated that our algorithm achieved a detection accuracy of 0.870 and a sensitivity of 0.901. Moreover, the method had a specificity of 0.807 and a precision of 0.88 for this same dataset. Our approach maintains good stability of physiological properties during deep learning, making it as robust as some single-class deepfake detection methods. The results of the tests on the selected datasets demonstrate higher accuracy compared to other methods.

Nowadays, the phenomenon of Deepfake is crucial as it enables the creation of extremely realistic images capable of deceiving anyone, thanks to deep learning tools based on Generative Adversarial Networks (GANs). These images are used as profile pictures on social media with the intent to sow discord and perpetrate scams on a global scale. In this study, we demonstrate that these images can be identified through various imperfections present in the synthesized eyes, such as the irregular shape of the pupil and the difference between the corneal reflections of the two eyes. These defects result from the absence of physical and physiological constraints in most GAN models. We have developed a two-level architecture capable of detecting these fake images. This approach begins with an automatic segmentation method for the pupils to verify their shape, as real image pupils naturally have a regular shape, typically round. Next, for all images where the pupils are not regular, the entire image is analyzed to verify the reflections. This step involves passing the facial image through an architecture that extracts and compares the specular reflections of the corneas of the two eyes, assuming that the eyes of real people observing a light source should reflect the same thing. Our experiments with a large dataset of real images from the Flickr-FacesHQ and CelebA datasets, as well as fake images from StyleGAN2 and ProGAN, show the effectiveness of our method. Our approach maintains good stability of physiological properties during deep learning, making it as robust as some single-class deepfake detection methods. The results of the tests on the selected datasets demonstrate higher accuracy compared to other methods.