uulmMAD - A Human Action Recognition Dataset for Ground-Truth Evaluation and Investigation of View Invariances.
Michael GlodekGeorg LayherFelix HeilemannFlorian GawrilowiczGünther PalmFriedhelm SchwenkerHeiko NeumannPublished in: MPRSS (2014)
Keyphrases
- action recognition
- human actions
- ground truth
- static images
- human movements
- human activities
- bag of words
- ucf sports
- video dataset
- activity recognition
- human subjects
- space time interest points
- spatial temporal
- action recognition in videos
- action classification
- computer vision
- human object interactions
- motion history images
- motion capture data
- human detection
- view invariant
- recognizing human actions
- bag of features
- recognition of human actions
- depth sensors
- body parts
- recognizing actions
- motion recognition
- space time
- view invariant action recognition
- action primitives
- action detection
- independent subspace analysis