Exploring Coarse-to-Fine Action Token Localization and Interaction for Fine-grained Video Action Recognition.
Baoli SunXinchen YeZhihui WangHaojie LiZhiyong WangPublished in: ACM Multimedia (2023)
Keyphrases
- fine grained
- coarse to fine
- action recognition
- human actions
- action classification
- recognizing human actions
- action detection
- coarse grained
- multiscale
- space time interest points
- recognition of human actions
- multiresolution
- object detection
- spatial temporal
- recognizing actions
- spatio temporal interest points
- bag of words
- activity recognition
- computer vision
- video dataset
- motion features
- human activities
- spatio temporal
- view invariant
- body parts
- action primitives
- bag of features
- access control
- video sequences
- image registration
- human motion
- space time
- human computer interaction
- human pose
- motion history images
- dynamic programming
- static images
- atomic actions
- video data
- visual features
- object segmentation
- multi class
- object recognition
- image segmentation