UNIMEMnet: Learning long-term motion and appearance dynamics for video prediction with a unified memory network.
Kuai DaiXutao LiChuyao LuoWuqiao ChenYunming YeShanshan FengPublished in: Neural Networks (2023)
Keyphrases
- long term
- recurrent networks
- learning process
- multimedia
- visual data
- prediction accuracy
- space time
- radial basis function network
- spatial and temporal
- learning algorithm
- connectionist networks
- dynamic scenes
- video streams
- spatio temporal
- short term
- peer to peer
- key frames
- video sequences
- dynamic textures
- object motion
- video objects
- reinforcement learning
- appearance features
- social networks
- neural network