Video face forgery detection via facial motion-assisted capturing dense optical flow truncation
Article
Article Title | Video face forgery detection via facial motion-assisted capturing dense optical flow truncation |
---|---|
ERA Journal ID | 18149 |
Article Category | Article |
Authors | Yang, Gaoming, Xu, Kun, Fang, Xianjin and Zhang, Ji |
Journal Title | The Visual Computer |
Journal Citation | 39 (11), pp. 5589-5608 |
Number of Pages | 20 |
Year | 2023 |
Publisher | Springer |
Place of Publication | Germany |
ISSN | 0178-2789 |
1432-2315 | |
Digital Object Identifier (DOI) | https://doi.org/10.1007/s00371-022-02683-z |
Web Address (URL) | https://link.springer.com/article/10.1007/s00371-022-02683-z |
Abstract | Deep learning advancements have resulted in breakthroughs in facial forgery techniques. Facial forgery videos are growing increasingly lifelike, making it impossible for people to tell the difference between the real and the fake. The proliferation of facial forgery techniques, as well as the slowness with which they may be detected, could jeopardize personal data security. As a result, it’s critical to look at approaches that can be trained on both real and fake videos and then utilized to identify facial forgeries or not in videos. This study found that forged face videos undergo truncation between consecutive frames of optical flow imaging after dense optical flow processing. We present an approach for detecting video face forgery that extracts and analyzes characteristics from real and fake material, then use those features to train classification models on Celeb-DF and FaceForensics++. In addition, we employ a unique facial double-triangle region to assist in the extraction of video inter-frame feature data. Experiments results show that the facial motion features extracted from the double triangle region successfully assist in capturing the dense optical flow truncation. Extensive evaluation suggests that our proposed approach is effective for video face forgery detection. |
Keywords | Deep learning; Face forgery detection ; Deepfakes detection ; Dense optical flow |
ANZSRC Field of Research 2020 | 460299. Artificial intelligence not elsewhere classified |
Public Notes | Files associated with this item cannot be displayed due to copyright restrictions. |
Byline Affiliations | Anhui University of Science and Technology, China |
School of Mathematics, Physics and Computing |
https://research.usq.edu.au/item/z02z1/video-face-forgery-detection-via-facial-motion-assisted-capturing-dense-optical-flow-truncation
50
total views1
total downloads5
views this month0
downloads this month