Model-Based Viewpoint Invariant Human Activity Recognition from Uncalibrated Monocular Video Sequence.
Zaw Zaw HtikeSimon EgertonYe Chow KuangPublished in: Australasian Conference on Artificial Intelligence (2010)
Keyphrases
- human activity recognition
- viewpoint invariant
- image sequences
- video sequences
- activity recognition
- human activities
- wide baseline
- object tracking
- camera motion
- computer vision
- dynamic scenes
- object class
- feature detectors
- feature points
- space time
- video data
- flow field
- motion segmentation
- ego motion
- structure from motion
- motion estimation
- spatio temporal
- object recognition
- pose estimation
- human motion
- frame rate
- optical flow
- viewpoint
- moving objects
- semi supervised
- epipolar geometry
- feature space
- training data