UltraGuard watermarks face videos in real time using ultrasound biometrics
Original title: Ultrasound-Assisted Biometric Watermarking Against Face Video Deepfakes
Smartphone ultrasound and video motion are fused into watermarks meant to survive forgery and editing.
The authors propose UltraGuard, a biometric watermark for facial videos intended to resist deepfake forgery and editing. A smartphone speaker and microphone emit ultrasound and capture biometric features, which are fused with spatio-temporal video features and embedded in real time. They report experiments showing resistance to forgery and editing at the source and during propagation. Deployment on users' mobile devices is described as future work.
Unconfirmed: Defense effectiveness is claimed by the authors and is not independently verified.
- The method uses the phone speaker and microphone, with no extra sensor described.
- Watermarks are content-aware and continue through transmission and propagation.
- Authors report defense against forgery and editing at the source and in transit.
- Mobile deployment is proposed for the future, not reported as shipped.
From the original article
1 min read at arxiv.orgCurrently, artificial intelligence-generated content (AIGC) poses a severe threat to facial video deepfakes, and many AIGC companies employ video watermarking to mitigate unauthorized fake video usage, but its effectiveness is limited due to watermark tampering and private-model forgery. Additionally, some schemes actively add watermarks to user videos.
By Guodong Liu · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.