Not made for each other- Audio-Visual Dissonance-based Deepfake Detection and Localization

Keyphrases

• Audio 50%
• Audio Channel 50%
• Audio Modality 50%
• Audiovisual Dissonance 100%
• Contrastive Loss 50%
• Cross-entropy Loss 50%
• Deepfake 50%
• Deepfake Detection 100%
• Deepfake Videos 50%
• Discriminative Features 50%
• Dissimilarity Measure 50%
• Dissonance 100%
• Dissonance-based 100%
• Facial Movement 50%
• Image Tampering Localization 50%
• Lip Movement 50%
• Lip Synchronization 50%
• TIMIT 50%
• Unnatural 50%
• Video Segment 50%
• Video-based 50%
• Visual Channels 50%
• Visual Modality 50%

Computer Science

• Contrastive Loss 33%
• Deepfakes 100%
• Discriminative Feature 33%
• Individual Modality 33%
• Visual Modality 33%