K-Tangent Spaces on Riemannian Manifolds for Improved Pedestrian Detection.
Andres SaninConrad SandersonMehrtash Tafazzoli HarandiBrian C. LovellPublished in: CoRR (2014)
Keyphrases
- pedestrian detection
- riemannian manifolds
- human body
- euclidean space
- tangent space
- object detection
- detection rate
- mean shift
- manifold learning
- geodesic distance
- multi class classification
- benchmark datasets
- vector space
- parameter space
- geometric structure
- feature space
- low dimensional
- covariance matrices
- three dimensional
- principal component analysis
- active learning
- reproducing kernel hilbert space
- object recognition