SE_EDNet: A Robust Manipulated Faces Detection Algorithm

Peng, Chaoyang; Yao, Lihong; Sun, Tanfeng; Jiang, Xinghao; Mi, Zhongjie

doi:10.1007/978-3-030-89029-2_6

Cited by 1 publication

(4 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…So we conduct experiments on it. Due to the data imbalance, 60 and 8 are set as the sampling rates for real and manipulated faces respectively, which are set according to SE_EDNet [14]. Table 3 gives the comparison with previous methods.…”

Section: Evaluation On Celeb-dfmentioning

confidence: 99%

“…Triplet [39] uses a triplet network architecture to detect Deepfakes. SE_EDNet [14] use Euclidean distance to reflect the similarity between vectors, and a new calculation method of attention mechanism is proposed. EfficientNet-B4 (Eff-B4) [35] is popular in the DeepFake Detection Challenge due to its performance.…”

Section: Evaluation On Celeb-dfmentioning

confidence: 99%

“…Method AUC score (%) Capsule [36] 93.2 I3D [33] 97.6 Triplet [39] 99.2 SE_EDNet [14] 99.7 Eff-B4 [35] 99.8 Ours 99.9…”

Section: Cross-dataset Evaluation On Dfdmentioning

confidence: 99%

“…ey regard deepfake detection as an image classification problem and use CNN to extract features. However, some image post-processing methods will reduce the performance [9,14,15] in the RGB domain, such as Gaussian noise, JPEG compression, and median filtering. In the frequency domain, previous work has used filters to preprocess the images.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Manipulated Faces Detection with Adaptive Filter

Peng

Sun

et al. 2022

Security and Communication Networks

View full text Add to dashboard Cite

With the progress of face manipulation techniques, synthesized faces are spreading on the Internet, which raises concerns about potential threats. To prevent these techniques’ abuse, various detection algorithms have been proposed. In this paper, we consider the image’s frequency information, then propose an adaptive filtering algorithm named spatial and adaptive filtering (SAF) Network. SAF is a dual-stream network that considers spatial and frequency domains. In the frequency domain, wavelet transform is used to divide the image into different frequency bands, then an adaptive filter is introduced, which aims to capture more decisive information by giving different weights to different frequencies. To fuse spatial and frequency features, spatial pyramid pooling fusion (SPPF) is proposed, which solves the mismatch of feature maps, and considers the relationship between different patches by attention mechanism. Experiment results show that the performance of SAF is better than the comparison algorithm.

show abstract

Section: Evaluation On Celeb-dfmentioning

confidence: 99%

Section: Evaluation On Celeb-dfmentioning

confidence: 99%

“…Method AUC score (%) Capsule [36] 93.2 I3D [33] 97.6 Triplet [39] 99.2 SE_EDNet [14] 99.7 Eff-B4 [35] 99.8 Ours 99.9…”

Section: Cross-dataset Evaluation On Dfdmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations