Abstract
The recent advance of synthetic image generation and manipulation methods allows us to generate synthetic face images close to real images. On the other hand, the importance of identifying the synthetic face images increases more and more to protect personal privacy from those. Although some deep learning-based image forensic methods have been developed recently, it is still challenging to distinguish synthetic images generated by recent image generation and manipulation methods such as the deep fake, face2face, and face swap. To resolve this challenge, we propose a novel generative adversarial ensemble learning method. We train multiple discriminative and generative networks based on the adversarial learning. Compared to the conventional adversarial learning, our method is however more focused on improving the discrimination ability rather than image generation one. To this end, we improve the discriminabilty by ensembling outputs from different two discriminators. In addition, we train two generators in order to generate general and hard synthetic images. By ensemble learning of all the generators and discriminators, we improve the discriminators by using the generated synthetic face images, and improve the generators by passing the combined feedback of the discriminators. On the FaceForensics benchmark challenge, we thoroughly evaluate our methods by comparing the recent methods. We also provide the ablation study to prove the effectiveness and usefulness of our method.
Highlights
The face plays an important role in confirming a person’s identity and understanding between human interactions [1]
Face images are considered as important cues in computer vision and machine learning areas, and many progress has been made in face detection, face recognition, and facial emotion recognition over the last decades
In this paper, we have proposed the generative adversarial ensemble learning for improving discriminability of manipulated face images
Summary
The face plays an important role in confirming a person’s identity and understanding between human interactions [1]. Fake image detection methods [7]–[9] based on deep learning have been flourished. To resolve this challenge, we propose a novel generative adversarial ensemble learning method. We train multiple discriminators and generators together. A goal of conventional adversarial learning methods is to improve the generation ability. We train the generators in different manners to improve the diversity of fake images. Other one is trained in an adversarial manner with ensembled discriminators As a result, this one can be trained to produce hard fake samples with the combined feedbacks of the discriminators. On the FaceForensics challenge benchmark dataset [10], we train and evaluate our generative adversarial ensemble method. Proposition of the adversarial ensemble method for improving discrimination ability;. Proposition of training strategies to make generators generate various and hard negative fake images
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
