Learning motion-difference features using Gaussian restricted Boltzmann machines for efficient human action recognition.
Son N. TranEmmanouil BenetosArtur S. d'Avila GarcezPublished in: IJCNN (2014)
Keyphrases
- action recognition
- static images
- human movements
- restricted boltzmann machine
- deep learning
- recognizing human actions
- motion history images
- active learning
- human actions
- recognition of human actions
- learning algorithm
- prior knowledge
- reinforcement learning
- image sequences
- feature vectors
- activity recognition
- action recognition in videos
- probabilistic graphical models
- human pose
- human motion
- bag of words
- space time
- feature space
- bayesian networks
- computer vision