AIM: Adapting Image Models for Efficient Video Action Recognition.
Taojiannan YangYi ZhuYusheng XieAston ZhangChen ChenMu LiPublished in: ICLR (2023)
Keyphrases
- action recognition
- static images
- human actions
- mid level
- action classification
- bag of features
- spatial temporal
- activity recognition
- video dataset
- input image
- human activities
- recognizing human actions
- image content
- computer vision
- action detection
- human detection
- multiscale
- recognition of human actions
- bag of words
- video sequences
- random fields
- video images
- image features
- motion features
- image classification
- image retrieval
- video data
- probabilistic model
- low level descriptors
- video surveillance
- motion history images
- human pose
- image representation
- video content
- d scene
- event detection
- key frames
- object categories