Dual Contrastive Learning for General Face Forgery Detection

Sun, Kening; Yao, Taiping; Chen, Shen; Ding, Shouhong; Jilin, L; Ji, Rongrong

doi:10.48550/arxiv.2112.13522

Cited by 4 publications

(5 citation statements)

References 16 publications

(10 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Evaluation Metrics. Following the previous works [9], [28], [57], we use Accuracy score (ACC), Area Under the Receiver Operating Characteristic Curve (AUC), and Equal Error Rate (EER) as our evaluation metrics. Data Preprocessing.…”

Section: Methodsmentioning

confidence: 99%

“…Considering the more challenging cross-dataset setting, we further evaluate our method on DFD, DFDCP, Deepwild, and CDF. Following the setup in the SOTA method [9], we train our model on the c23 version of FF++ and test on other four datasets. The results are shown in Table III, where our method presents superior performance on all four datasets.…”

Section: A Comparing With Previous Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Detecting Deepfake by Creating Spatio-Temporal Regularity Disruption

Guan¹,

Zhou²,

Gong³

et al. 2022

Preprint

View full text Add to dashboard Cite

Despite encouraging progress in deepfake detection, generalization to unseen forgery types remains a significant challenge due to the limited forgery clues explored during training. In contrast, we notice a common phenomenon in deepfake: fake video creation inevitably disrupts the statistical regularity in original videos. Inspired by this observation, we propose to boost the generalization of deepfake detection by distinguishing the "regularity disruption" that does not appear in real videos. Specifically, by carefully examining the spatial and temporal properties, we propose to disrupt a real video through a Pseudo-fake Generator and create a wide range of pseudo-fake videos for training. Such practice allows us to achieve deepfake detection without using fake videos and improves the generalization ability in a simple and efficient manner. To jointly capture the spatial and temporal disruptions, we propose a Spatio-Temporal Enhancement block to learn the regularity disruption across space and time on our self-created videos. Through comprehensive experiments, our method exhibits excellent performance on several datasets.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: A Comparing With Previous Methodsmentioning

confidence: 99%

Detecting Deepfake by Creating Spatio-Temporal Regularity Disruption

Guan¹,

Zhou²,

Gong³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Detection of facial manipulation, as a passive defense measure against abuses, has been widely studied in recent years. Most works (Sun et al 2021b;Qian et al 2020;Li et al 2020;Guan et al 2023c) are formulated as a binary classification problem, with the objective of identifying preexisting fake contents. While powerful deep neural networks could provide a good understanding of the differences between real faces and the generated fake ones, recent studies (Sun et al 2021a;Luo et al 2021;Guan et al 2022;Dong et al 2022;Yao et al 2023;Yan et al 2023;Dong et al 2023) find it hard to keep a consistent performance from the training set to untapped manipulations.…”

Section: Facial Manipulation Detectionmentioning

confidence: 99%

Adversarial Robust Safeguard for Evading Deep Facial Manipulation

Guan,

Zhao,

et al. 2024

AAAI

View full text Add to dashboard Cite

The non-consensual exploitation of facial manipulation has emerged as a pressing societal concern. In tandem with the identification of such fake content, recent research endeavors have advocated countering manipulation techniques through proactive interventions, specifically the incorporation of adversarial noise to impede the manipulation in advance. Nevertheless, with insufficient consideration of robustness, we show that current methods falter in providing protection after simple perturbations, e.g., blur. In addition, traditional optimization-based methods face limitations in scalability as they struggle to accommodate the substantial expansion of data volume, a consequence of the time-intensive iterative pipeline. To solve these challenges, we propose a learning-based model, Adversarial Robust Safeguard (ARS), to generate desirable protection noise in a single forward process, concurrently exhibiting a heightened resistance against prevalent perturbations. Specifically, our method involves a two-way protection design, characterized by a basic protection component responsible for generating efficacious noise features, coupled with robust protection for further enhancement. In robust protection, we first fuse image features with spatially duplicated noise embedding, thereby accounting for inherent information redundancy. Subsequently, a combination comprising a differentiable perturbation module and an adversarial network is devised to simulate potential information degradation during the training process. To evaluate it, we conduct experiments on four manipulation methods and compare recent works comprehensively. The results of our method exhibit good visual effects with pronounced robustness against varied perturbations at different levels.

show abstract

“…These methods only utilize the information from the spatial domain, which generally overfits the classification boundary. On the other hand, some works [11,28,30,36,40,49] observe the diversity of real faces and fake faces in the frequency domain and propose the face forgery detection method with the frequency clues. F 3 -Net [36] integrates the frequency-aware decomposition and local frequency statistics into a whole learning framework to classify the real and fake faces.…”

Section: Face Forgery Detectionmentioning

confidence: 99%

Exploring Frequency Adversarial Attacks for Face Forgery Detection

Jia¹,

Ma²,

Yao³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Various facial manipulation techniques have drawn serious public concerns in morality, security, and privacy. Although existing face forgery classifiers achieve promising performance on detecting fake images, these methods are vulnerable to adversarial examples with injected imperceptible perturbations on the pixels. Meanwhile, many face forgery detectors always utilize the frequency diversity between real and fake faces as a crucial clue. In this paper, instead of injecting adversarial perturbations into the spatial domain, we propose a frequency adversarial attack method against face forgery detectors. Concretely, we apply discrete cosine transform (DCT) on the input images and introduce a fusion module to capture the salient region of adversary in the frequency domain. Compared with existing adversarial attacks (e.g. FGSM, PGD) in the spatial domain, our method is more imperceptible to human observers and does not degrade the visual quality of the original images. Moreover, inspired by the idea of meta-learning, we also propose a hybrid adversarial attack that performs attacks in both the spatial and frequency domains. Extensive experiments indicate that the proposed method fools not only the spatial-based detectors but also the state-ofthe-art frequency-based detectors effectively. In addition, the proposed frequency attack enhances the transferability across face forgery detectors as black-box attacks.

show abstract

Dual Contrastive Learning for General Face Forgery Detection

Cited by 4 publications

References 16 publications

Detecting Deepfake by Creating Spatio-Temporal Regularity Disruption

Detecting Deepfake by Creating Spatio-Temporal Regularity Disruption

Adversarial Robust Safeguard for Evading Deep Facial Manipulation

Exploring Frequency Adversarial Attacks for Face Forgery Detection

Contact Info

Product

Resources

About