Sequence of the most informative joints (SMIJ): A new representation for human skeletal action recognition.
Ferda OfliRizwan ChaudhryGregorij KurilloRené VidalRuzena BajcsyPublished in: J. Vis. Commun. Image Represent. (2014)
Keyphrases
- action recognition
- recognizing human actions
- spatio temporal interest points
- recognition of human actions
- body parts
- view invariant
- human actions
- action recognition in videos
- static images
- human motion
- human movements
- human activities
- activity recognition
- bag of features
- bag of words
- motion history images
- human detection
- computer vision
- motion capture data
- spatial temporal
- action classification
- human object interactions
- temporal structure
- human pose
- atomic actions
- recognizing actions
- image representation
- depth sensors
- image sequences
- multiscale
- action primitives
- motion primitives
- video dataset
- motion capture
- object recognition
- machine learning
- action detection
- image classification
- independent subspace analysis