DKD-DAD: a novel framework with discriminative kinematic descriptor and deep attention-pooled descriptor for action recognition.
Ming TongMingyang LiHe BaiLei MaMengao ZhaoPublished in: Neural Comput. Appl. (2020)
Keyphrases
- action recognition
- action classification
- recognizing actions
- bag of words
- video dataset
- activity recognition
- max margin
- human actions
- recognition of human actions
- computer vision
- mid level
- action recognition in videos
- human detection
- action detection
- recognizing human actions
- bag of features
- spatial temporal
- body parts
- human activities
- spatio temporal interest points
- action primitives
- human pose
- deformable part models
- static images
- human computer interaction
- face recognition
- view invariant
- discriminative power
- semi supervised