MSG-BART: Multi-granularity Scene Graph-Enhanced Encoder-Decoder Language Model for Video-grounded Dialogue Generation.
Hongcheng LiuZhe ChenHui LiPingjie WangYanfeng WangYu WangPublished in: CoRR (2023)
Keyphrases
- language model
- multi granularity
- video sequences
- language modeling
- bit budget
- video codec
- motion estimation
- n gram
- distributed video coding
- low complexity
- video encoder
- probabilistic model
- rate distortion
- multi user
- video data
- information retrieval
- dynamic integration
- query expansion
- video content
- smoothing methods
- bit rate
- multimedia
- video frames
- ad hoc information retrieval
- real time
- test collection
- image sequences
- retrieval model
- moving objects
- translation model
- bitstream
- mixture model
- weighted graph
- privacy protection
- location aware
- wyner ziv
- video coding
- query processing
- bayesian networks
- virtual world
- motion vectors