Generative Adversarial Ensemble Learning for Face Forensics

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. INDEX TERMS Digital image forensics, generative adversarial ensemble learning, deep learning, synthetic image detection, face image.

Unveiling Spoofing Attempts: A DCGAN-based Approach to Enhance Face Spoof Detection in Biometric Authentication

“…Although there are certain restrictions in domain adaptation strategies, Kumar et al [13] focus on metric classification and triplet network design emphasises the significance of feature space differentiation. The Generative Adversarial Ensemble Learning method by Baek et al [14] prioritises discrimination enhancement and achieves respectable accuracy at the cost of training time and computational resources. The CNN model developed by Ranjan et al [15] using Transfer Learning exhibits better performance on various datasets; nonetheless, its dependence on pre-trained models could result in less-than-ideal outcomes.…”

“…[13] [14] [15]. Baek et al [14] proposed Generative Adversarial Ensemble Learning for Face Forensics which involves multiple discriminative and generative networks. Unlike conventional approaches, it focuses on enhancing discrimination rather than image generate on.…”

Countering Spoof: Towards Detecting Deepfake with Multidimensional Biological Signals

“…We hope that the deepfake detection in these scenarios could be solved effectively by using signal enhancement and denoising in the near future work. [27] 84.5 73.7 Rahmouni et al [28] 85.5 64.2 Baek et al [29] 71.8 68.6 Rossler et al [30] 96.4 86.9 Dogonadze [31] 93.6 83.9 Ours 98.01 93.94…”

Lite-3DCNN Combined with Attention Mechanism for Complex Human Movement Recognition

“…Similarly, Wang and Dantcheva [11] trained and finetuned 3D ResNet [12] on the well-known FaceForensic++ dataset, which is an excellent motion recognition network [13]. In addition, generative adversarial networks (GAEL Net [14]) have also been used to design robust facial manipulation detectors. erefore, researchers began designing more complex architectures to achieve higher detection accuracy.…”