Exposing AI-generated videos with motion magnification
- PDF / 2,629,996 Bytes
- 14 Pages / 439.37 x 666.142 pts Page_size
- 80 Downloads / 252 Views
Exposing AI-generated videos with motion magnification Jianwei Fei 1 & Zhihua Xia 1 & Peipeng Yu 1 & Fengjun Xiao 2 Received: 23 January 2020 / Revised: 6 April 2020 / Accepted: 27 May 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
Recent progress of artificial intelligence makes it easier to edit facial movements in videos or create face substitutions, bringing new challenges to anti-fake-faces techniques. Although multimedia forensics provides many detection algorithms from a traditional point of view, it is increasingly hard to discriminate the fake videos from real ones while they become more sophisticated and plausible with updated forgery technologies. In this paper, we introduce a motion discrepancy based method that can effectively differentiate AI-generated fake videos from real ones. The amplitude of face motions in videos is first magnified, and fake videos will show more serious distortion or flicker than the pristine videos. We pre-trained a deep CNN on frames extracted from the training videos and the output vectors of the frame sequences are used as input of an LSTM at secondary training stage. Our approach is evaluated over a large fake video dataset Faceforensics++ produced by various advanced generation technologies, it shows superior performance contrasted to existing pixel-based fake video forensics approaches. Keywords Deep learning . Fake videos . DeepFakes detection . Motion magnification
* Zhihua Xia [email protected] Jianwei Fei [email protected] Peipeng Yu [email protected] Fengjun Xiao [email protected]
1
Jiangsu Engineering Center of Network Monitoring, Jiangsu Collaborative Innovation Center on Atmospheric Environment and Equipment Technology, School of Computer and Software, Nanjing University of Information Science & Technology, Nanjing 210044, China
2
Hangzhou Dianzi University, Management School, Hangzhou 310018, China
Multimedia Tools and Applications
1 Introduction The ever deeper participation of individuals in social networking and sharing portals leaves generous and direct access to personal videos. The illegal is thus able to get these videos in a breeze and forge a fake one with the help of multimedia generation and manipulation technology. These technologies are experiencing great advances with the help of deep learning and have earned a common name “AI-Face-Synthesis” technology which has raised a dramatic sense of crisis among academia and the authorities. They significantly reduce the cost in manufacture and manipulation of videos and do not require much technical knowledge timeconsuming process. Human faces are very important biological characteristic for authentication and have been marked by these technologies. Unlike other biological information such as iris or fingerprints which are relatively easy to be verified for authenticity and hard to be tampered [10, 11, 36, 45], human faces in videos are easier to be manipulated. Led by DeepFakes [14], these technologies synthesize fake videos in the form of: 1) face-swapping tha
Data Loading...
