Maximizing Spatio-Temporal Entropy of Deep 3D CNNs for Efficient Video Recognition.
Junyan WangZhenhong SunYichen QianDong GongXiuyu SunMing LinMaurice PagnuccoYang SongPublished in: ICLR (2023)
Keyphrases
- spatio temporal
- spatial and temporal
- spatial temporal
- space time
- recognition rate
- object recognition
- human actions
- multimedia
- video frames
- video data
- action recognition
- video sequences
- recognition accuracy
- human activities
- information theory
- pattern recognition
- video representation
- video streams
- video shots
- recognition process
- video processing
- cellular neural networks
- video copy detection
- video analysis
- dynamic scenes
- recognition algorithm
- motion estimation
- moving objects
- feature extraction
- face recognition
- image sequences