Phenaki: Variable Length Video Generation from Open Domain Textual Descriptions.
Ruben VillegasMohammad BabaeizadehPieter-Jan KindermansHernan MoraldoHan ZhangMohammad Taghi SaffarSantiago CastroJulius KunzeDumitru ErhanPublished in: ICLR (2023)
Keyphrases
- variable length
- textual descriptions
- open domain
- fixed length
- semantic representation
- information extraction
- metadata
- semantic concepts
- n gram
- web images
- video data
- question answering
- multimedia
- bitstream
- video analysis
- video frames
- key frames
- semantic information
- question answering systems
- video content
- web search engines
- image data
- keywords
- bayesian networks