3D SMoSIFT: three-dimensional sparse motion scale invariant feature transform for activity recognition from RGB-D videos.
Jun WanQiuqi RuanWei LiGaoyun AnRuizhen ZhaoPublished in: J. Electronic Imaging (2014)
Keyphrases
- activity recognition
- motion features
- scale invariant feature transform
- video stabilization
- three dimensional
- depth images
- human activities
- human activity recognition
- keypoints
- accelerometer data
- action recognition
- camera motion
- feature points
- bag of features
- human actions
- sift descriptors
- image sequences
- object recognition
- image matching
- pose estimation
- depth map
- feature matching
- video processing
- high dimensional
- motion estimation
- multi view
- dynamic scenes
- human body
- d objects
- video sequences
- human motion
- visual features
- dynamic textures
- feature descriptors
- shape features
- image pairs
- key frames
- sparse representation
- machine learning