Multi-Definition Video Deepfake Detection via Semantics Reduction and Cross-Domain Training.
Chutian WangCairong ZhaoGuosheng HuPublished in: ICME (2022)
Keyphrases
- cross domain
- domain adaptation
- video data
- video content
- transfer learning
- video sequences
- knowledge transfer
- video frames
- sentiment classification
- text categorization
- multiple domains
- cross domain learning
- event detection
- semantic information
- supervised learning
- object detection
- target domain
- video retrieval
- active learning
- training set
- event recognition
- natural language
- web image annotation
- case study