Event
RoboFake Test Detector's Defense Against Humanoid Robot Face Attacks
1 reports · 1 sources
Hao Chen and Junnan Xu propose the RoboFake attack: using a real humanoid facial robot to synchronize lip movements and expressions, combined with generated speech, to constitute a presentation attack without pixel editing, and tested three face presentation attack detection (PAD) models and six deepfake detectors. All results are compared with real human face videos that underwent the same processing and were turned to the same head angles. The paper states that current models can recognize the humanoid robot face because its face does not look human, rather than because the attack was identified.
Report timeline
- RoboFake: Can Deepfake Detectors Recognize Humanoid Face Robots?
The RoboFake paper proposes using a real humanoid face robot with synchronized lip movements and expressions, combined with generated speech, to create a new type of presentation attack with no pixel editing, and tests three public presentation attack detection (PAD) models and six deepfake detectors.
arXiv cs.RO ·