AdaFocus V2: End-to-End Training of Spatial Dynamic Networks for Video Recognition.
Yulin WangYang YueYuanze LinHaojun JiangZihang LaiVictor KulikovNikita OrlovHumphrey ShiGao HuangPublished in: CoRR (2021)
Keyphrases
- end to end
- dynamic networks
- scalable video
- network structure
- temporal correlation
- video data
- congestion control
- multimedia
- video sequences
- temporal evolution
- video content
- admission control
- spatial information
- network analysis
- text localization and recognition
- video frames
- dynamic behavior
- video streams
- training set
- ad hoc networks
- temporal information
- real world networks
- data mining
- real time
- bitstream
- compressed video
- motion vectors
- knowledge discovery
- transport layer